Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=

# When Competition Compels Change: Trade, Management, and Productivity

Authors:   
Discussant: None  
Video: https://www.youtube.com/watch?v=PHYYOyrSItw&t=5429s


## Talk (01:30:29 – 02:23:10)

[01:30:29] Take All right. If you could please please  
[01:30:59] take your seats so that COIA poor COIA doesn't get 42 minutes.  
[01:31:08] >> Is this working? All right, folks. Let's sit down.  
[01:31:46] Okay everyone we are happy to start the next session. Um we have Kotia here first telling us about competition uh again um in trade and management.  
[01:31:58] >> Thanks very much uh Sundep. Uh thanks very much for including me in the program. Uh I'm going to be presenting what was my job market paper last fall when competition compels change trade management and productivity.  
[01:32:08] >> [gasps] >> So to motivate the project, let me begin by this quote by Helman and Krugman which says that the idea that international trade increases competition goes back to Adam Smith and that this is one of the reasons why economists believe that gains from trade and the costs of protection are larger  
[01:32:23] than what their own models suggest. Now in the last three decades we've made great uh progress in understanding this link between trade competition and productivity with canonical models emphasizing the idea that trade drives  
[01:32:36] out lagard firms and that this tougher selection leads to productivity gains at the industry level. Now the focus of such work is decidedly external to the firm in the sense that these gains are coming from the reallocation of  
[01:32:49] resources between firm. What I'm going to do in this project is uh to show that trade can additionally also reduce within firm slack or x inefficiency which is this idea that absent competitive pressures firms may not  
[01:33:03] optimally allocate their resources. And so more concretely what I do is to show that in an environment where domestic firms are protected from foreign competition um firms may not have the incentive to invest in managerial  
[01:33:16] quality but some of the very same firms might be compelled by competition to invest in management quality and firm organization and I'll be explicit about these two terms in the next slide. um and that might be accompanied with uh  
[01:33:30] productivity gains within the firm and might be correlated with productivity gains. And in this uh global context that we face today where we're seeing a retreat from free trade and a renewed emphasis emphasis on um protectionism, I  
[01:33:43] think understanding how trade shapes organizations has gained renewed urgency. So what do I mean by firm organization and management quality and for that look at this picture that represents uh the top executives of an  
[01:33:55] Indian firm. It's not the most diverse picture. You see you see eight men here, but if you look closely, they're actually all called George.  
[01:34:06] And this is this is taken from this firm's uh from this firm's website. And this is this is not a coincidence. Uh George is a common family name in the southern union state of Kerala. And they're indeed all related to each other. And this is also not an  
[01:34:20] exception. Firms like this where the top managerial talent comes from a narrow pool of one founders's family are very common in India and to some extent represent the predominant form of corporate governance uh across the development spectrum both in rich and  
[01:34:33] poor countries. Um and if you think that talent is more widely dispersed in the economy then firm organization like this probably represents poor allocation of managerial talent that can depress firm productivity.  
[01:34:45] Now, of course, there can be many ways why firms choose to organize this way.  
[01:34:49] Uh, and I can talk about that u later on in the talk, but one that I'm going to focus on because this really came um, you know, was the most predominant reason that firms mentioned to me when I spoke to some of them was this idea of nonmonetary private benefits. This idea  
[01:35:04] that I like to organize my firm around the family that gives me some satisfaction. And there's also a long tradition in the corporate finance literature thinking about family firms as originating partly because of this reason. And so what I do in this project  
[01:35:16] is to show that trade and competition can induce firms like this to professionalize their management. And the key mechanism is that competition which is a bad shock. It reduces my profits. uh eliminates the luxury of  
[01:35:30] prioritizing these non-pukunity private benefits and really focus on the bottom line uh whether I can stay alive and have positive profits or not and I think this helps us understand more holistically the development process and  
[01:35:42] particularly the role of market forces uh and competition in it because these organizational changes that I'm going to document are not happening at a time that India is going through any deeper cultural legal institutional changes and so I think uh that's helpful from a from  
[01:35:56] understanding development as well. So I do this really in three steps. Uh in step one I obtain uh first time access to confidential administrative data on family type and tenure histories of over 6 million company directors who work on  
[01:36:11] the universe of Indian registered firms allowing me a unique glimpse to identify and quantify how much families there within these firms and how that changes over time and second to understand how these firms respond to a large import  
[01:36:25] liberalization episode. uh I exploit uh a natural experiment uh which is product specific and it therefore lends itself to an event study design and there's really three results that I want to focus on. The first is that this is a  
[01:36:38] bad shock for firms. Their profits are going down as they're competing with foreign goods. Uh and I I document that firm revenues, firm profits, they decline quite substantially and persistently. Uh revenues down by 28%  
[01:36:51] over an 8-year period. How do then firms respond to this negative shock? Family firms actually start professionalizing their management by which I mean that they replace some of the family managers on the top. So these jes are getting  
[01:37:05] some outside talent now into their firm. Uh so the share of family executives in the firm declines by about 15% relative to the pre-policy control group mean.  
[01:37:14] And what's really crucial is that almost all of this professionalizing behavior is driven by lagard firms. And this is very different therefore from a story where uh the frontier firms in the economy are induced by trade or export opportunities to invest in better  
[01:37:28] quality uh better technology or innovation. The action here is coming from a very different part of the productivity distribution and that's going to to be key uh both in the model and in how we interpret the results. Um and finally firms that do  
[01:37:41] professionalize their management report um quantity productivity or TSPQ gains of about 30%. Okay. And then in the third part of the project to understand aggregate implications, I'm going to develop a quantitative framework. Uh the  
[01:37:56] the key mechanism here is going to be that competition compels um low productivity firms to professionalize primarily as a survival mechanism. So I really need these productivity gains as a marginal firm after competition.  
[01:38:09] Otherwise, I'm going to have to exit. My profits are negative. Um and therefore I'm going to give up these private benefits and professionalize. And the aggregate implications of this firm level response uh they're actually not obvious in theory for a bunch of um you  
[01:38:23] know forces that go against each other. So on the one hand professionalizing firms becomes more uh they become more productive and so that's good for aggregate productivity but at the same time saving these lagards and preventing them from exit weakens the selection  
[01:38:36] mechanism uh which is uh which is going to be a drag on aggregate productivity because now the weakest firms are not being thrown out they're actually staying alive in the economy. But while saving these firms is bad for aggregate productivity, it might still be good for  
[01:38:50] welfare because consumers value variety. And so to some extent you're giving consumers access to more varieties and it might enhance uh welfare. And so to put numbers on these opposing forces, I'm going to have uh a quantification exercise so that we can understand the  
[01:39:05] aggregate effect. Great. Um I'm going to skip over the literature except to just mention the second point here. There's a lot of work on managerial practices and how they can induce within firm gains in productivity for firms who implement better  
[01:39:19] management practices. But I think the key question is if these gains are present uh why aren't firms already on their own implementing these better management practices. And I think to some extent what I show is that poor management uh uh practices might be  
[01:39:34] there just because of low product market competition. firms are not compelled therefore to implement these um uh you know and and realize these productivity gains.  
[01:39:44] Great. Um so I'm going to organize the talk as follows. First I'm going to go over the raw materials of the empirical results. That's the data and the natural experiment. Uh I'm going to then talk about uh event studies to understand firm level responses and finally a model  
[01:39:59] uh to put some numbers uh on these things. Yeah.  
[01:40:02] >> U thanks. The one uh one thing I'm a little puzzled about is that the word trust did not really appear in your introduction, right? And you were saying there's some non-pickinary benefits. I like my family running things, but it could just be in the absence of formal  
[01:40:16] institutions of within firm disputes, right? You're using this.  
[01:40:20] >> Absolutely. No, I I I talk about that in a lot of detail in the paper. I think the way I think about this is that um think about trust as a cost if I professionalize. So you could think of that as a wedge on profits in the state of the world where I professionalize.  
[01:40:33] And so that's a monetary cost that I'm paying. When I'm facing a negative shock which is reducing my profits, that's precisely the time when I'm less able to pay an additional cost, you know, because of say loss of expropriation, etc. And so in a world with only  
[01:40:47] contracting frictions without these non-monetary gains, it's difficult to get more professionalization. In a world where my profits are going down, you would actually predict that there'll be less professionalization. I'm going to add that in the model when I discuss that and and so you can maybe see it  
[01:41:01] more transparently but I think you you can't get even the qualitative prediction that you'll see more professionalization precisely when you're having a negative shock.  
[01:41:09] >> Yeah. Uh have you thought about some positive effects of private benefits in the sense that you know I'm prideful about my company so that gives me extra incentive to you know make a long-term investment etc. So maybe those things  
[01:41:23] might go away because of the shock. So there can be some negative effect. Oh, I see. So, um I think that's fine to the extent that you'll see in the way I model this, I have sort of this one parameter that  
[01:41:37] gets at the difference between how productive family versus non-family firms are. Um and what you're saying is to some extent will attenuate that parameter, right? So, I think I'm I'm okay with that. Uh I'm I'm okay with  
[01:41:49] that interpretation. Yeah. >> The long benefits could be learning. So, you don't know good your manager is and I think it's adversity that makes makes the point that you know this  
[01:42:03] guy is not good leads to both right >> that make any difference >> I think that's a great point so if you if you if your beliefs are being updated about the quality of your family versus professional management that that could um that could make you act precisely  
[01:42:17] when you face a shock um I'm not so sure if it would only make the lagards change uh unless the beliefs are worse of the lagards than the frontier firms. Um but also I actually find remarkably not that  
[01:42:32] much churn in professional firms. So to the extent that beliefs might be uh you know uh might be wrong for both professional and family firms in the face of a shock you should see that belief updating and management churned then in both professional and family  
[01:42:45] firms. But I sort of see it concentrated amongst family and within family only for the least productive family firms.  
[01:42:51] Uh but that's something that I could I could explore. Yeah.  
[01:42:55] If comes out to be a loser, that's going to only show up in in the loser.  
[01:43:01] It works precisely as results. I >> I agree. But I think what would if it was only a belief story without any private benefits. Um you would also see at least some churn in all firms that are already professional because I know  
[01:43:13] that my CEO is is not talented and the short makes me realize that my CEO is really untalented and I need to throw him out and get someone new. Um but but I could think about that uh in more detail.  
[01:43:25] Yes. >> Yes.  
[01:43:29] But one other equity or new shareholder which come with preconditions of changing the board.  
[01:43:38] >> Yeah. >> I was wondering if that >> so here I'm going to make a distinction between ownership which this paper is about and uh sorry management which this paper is about and ownership where I don't have good data. Um what I find for the years that I do have data, this is a  
[01:43:53] period where India is financially liberalizing. So you see a secular decline in the share of equity held by family owners that's happening both for firms who professionalize their management and for firms who don't professionalize their management. So I'm not able to say more than that on ownership. It's kind of this common  
[01:44:07] thing happening to both these firms, but in in a companion project with Nick Bloom and Pete Cleo, we're looking at exactly that at ownership and how that interacts with people who are on the board. Yeah. Okay. Great. So let me now talk about the natural experiment. Um  
[01:44:22] India has been one of the most closed economies since the 1950s. And one of the hallmarks of this regime was this idea of quantu quantitative restrictions. Think of these as almost complete import bans where I'm restricting quantitatively from goods  
[01:44:36] entering my economy. Um in 1991 there was a big IMF supported um structural adjustment program where India liberalized to some extent. But actually many of these quantitative restrictions  
[01:44:49] on an on as many as 3,000 goods defined at the HF8 digit level which represented 30% of all tariff lines. These remained in place even despite the first generation reform and the reason was that the WTO actually has a specific  
[01:45:02] clause saying that if you're a developing country which is actually self-declared and if you have fragile balance of payments you can put in these restrictions. And so what I'm going to do in this project is to exploit India's second generation trade reforms where  
[01:45:16] the US and the European Union took India to the WTO disputes resolution saying that now your external situation is actually quite comfortable. Uh and the IMF was asked to administer this independently and India lost the case was given about 3 years to remove all  
[01:45:30] these quantitative restrictions without any selection. Um and uh and make them free uh for imports. Okay, I'm not going to go into the details, but I show that imports of affected products in event studies uh rise quite substantially by  
[01:45:44] 30% relative to the control group. There's almost no impact on export. So, this is a unique shock and uniquely placed to ask the impact of competition on firm level responses because first um I have variation across the product  
[01:45:58] scope in which firms are affected and which firms are not. Second, this is externally imposed to some extent. it was um not an outcome of domestic political decisions that the Indian government took but the IMF actually has a rubric which decides whether a country  
[01:46:13] is or not uh fragile by its definition in terms of balance of payments are concerned. So it was externally imposed um and also unlike the first generation reforms in 1991 which were muddled with many many other domestic reforms domestic financial liberalization  
[01:46:27] industrial policy etc. Here these reforms are being implemented in relative isolation at least at this 8digit level and most crucially that there's no change in export policy and even within import competition it's  
[01:46:41] mostly final consumer goods that are facing competition. It's not intermediate product. So this is not a shock that primarily is making firms able to afford better quality inputs from abroad. This is really a shock where firms in their output markets are  
[01:46:54] facing competition because India's you know importing goods like textile, apparel, uh electronics etc. Okay, great. Uh so now I'm going to talk about the data construction uh for the project  
[01:47:08] and the first step of that is to try and identify the affected products uh for which I digitized thousands of these government of India notifications which tell you which product was liberalized in which year. And so here you see uh shrimp uh for example is being  
[01:47:23] liberalized and the policy column next to it is free. So I know the product descriptions. I also know product codes that I can then match to my firm level data to try and make a data set of product level import liberalization.  
[01:47:35] And this level of um identifying treatment is much sharper than previous studies in India or or even in Brazil where I have 10,000 such products. So the treatment is defined at that granularity. uh but past studies have  
[01:47:49] really uh defined treatment at broader two two digit uh two uh three-digit industry codes of which there's about 200.  
[01:47:57] The second step is to identify which firms are impacted by this product level import competition shock. And for that um unfortunately before AI because uh it didn't get better uh uh as it has now I had to do this by hand to match these  
[01:48:10] firm level descriptions with descriptions in the customs notifications that I just showed you to get a sense of which firms in my firm level data are impacted by this product level shock. And finally I use um the CMI prowess which is an Indian firm  
[01:48:25] level data data set. Think of it as large and medium-sized corporatized Indian firms. So these aren't small mom and pop establishments or tiny manufacturing plants. Think of firms with more than 30 40 employees. Um they  
[01:48:39] represent about 80% of corporate taxes collected in India. And what this data gives me is the product scope of the firm. So I know which products firms are producing. So I can identify them as treatment or control and also balance sheet as well as the financial  
[01:48:52] statements. And finally with these I combine uh to understand the impact on the managerial side data from the ministry of corporate affairs which gives me access to family ties and tenure histories of over 6 million company executives and to give you a  
[01:49:07] sense of what this data looks like um it's displayed for an anonymized firm on the right here where in the first column I know everyone on the board of directors uh so uh this is primarily a firm dominated by the Kosla family and then crucially I have a variable called  
[01:49:22] the father's name because of an idiosyncrasy in the way data are collected in India as a marker of identity along with your name you also mention your father's name and so through this one variable I can get a good sense of how much family there is on the board uh so you can see for  
[01:49:36] example that am and p kosla are all siblings because they report dk kosla as their father dk kosla himself happens to be on the board so there's multiple generation as is dk kosla's brother nk kosla because both of them report kla as  
[01:49:51] their father And so through this one variable, I get a sense of what share of the management of the firm is really coming from the founders's family. I'm going to focus on executive directors.  
[01:50:03] These are the CEOs, the managing directors, and not the independent directors. And I'm going to summarize one key number for each firm. What share of your executive directors are coming from the family? And I'm going to track that before and after competition. I  
[01:50:18] also have some interesting demographic data that I haven't done much with right now but I have uh stuff like uh you know the gender age as well as the place of birth uh which I think is an important marker of cast in India and and something could be done with that later.  
[01:50:30] Yeah. >> Um uh if you could help me understand the identifying variation a little bit more. So of course this trade liberization was a huge shock in India in the early 90s. Lots of things opened up. Um so I'm worried about whether you  
[01:50:43] know like so you have farms started also in importing intermediate goods for their production right so you could imagine another story where you know like if you're importing some intermediate goods or machines right you need  
[01:50:58] >> you know a different type of manager to to deal with that new style of doing business uh so is that a concern in this given your variation >> um so I think I've checked for that so so two responses on that first precisely ly because I'm not looking at that  
[01:51:11] broader Indian liberalization from 1991 and looking at something much later. I can to some extent rule out that intermediate goods um are are entering India at a faster pace than before. In the late 2000s, it was primarily consumer goods which were liberalized as  
[01:51:25] I said uh previously. So it's think of it as output market competition not a shock to the cost side of the firm. Um but in my data I know uh whether um a firm is changing its product scope on  
[01:51:39] the output side. So are they producing new products or not? Um and which is similar to whether they're using new inputs or not. And I actually find that they're not moving to new product lines.  
[01:51:48] So it's not that they need new expertise to produce new products. And I can do a similar exercise for inputs because I I think I have inputs separated by whether they're imported or domestically sourced. So I could do something like  
[01:52:00] that um here. Yeah. >> And there were huge effects. I think there was a study in the mountain.  
[01:52:12] >> Yeah. >> Which look at size.  
[01:52:15] >> So I'm wondering if you you look at some of those.  
[01:52:18] >> Yeah. So that was going to be my job market paper before I changed it. Uh so I have looked at that. Uh I have looked at that. Um that actually opened up quite a bit later than this policy.  
[01:52:28] happy to talk about that more but there is an overlap and to some extent uh you know yeah there are interesting uh interactions possible there yeah I can drop all the products which were liberalized for that policy and the results that I show still hold true okay  
[01:52:42] great so before going into the event studies let me talk about just what the pre-policy economy looks like in terms of firm size and profitability where I'm going to split the firms in red by firms who have a lot of family on their board so greater than 2/3 and then on the blue  
[01:52:56] firms that have very little family uh on their board and so you see this difference in size and profitability for firms in the baseline that I just want you to keep in mind uh you know as I as I show the event studies okay  
[01:53:09] so I'm going to estimate uh event studies of the following form where I'm going to identify a firm is treated if its highest revenue product is impacted uh by this QR removal and faces import liberalization um this is robust to  
[01:53:23] various alternative definitions um and I'm going to have firm fixed effects of course in these regressions but also because the shock is at the detailed 8digit level I'm able to put in broader sectorbyear fixed effects so think about this as uh uh you know looking at a  
[01:53:38] broader industry two different products one which was liberalized and one which was not and because the timing is staggered over three uh 3 to 5 years I'm going to use sun and Iraham uh to estimate these regressions but two-way fixed effects look almost identical  
[01:53:52] great so I'm going to show three sets of results the first is on firm contraction. Um and here you see that log revenue declines substantially and persistently after firms are facing import liberalization. These results  
[01:54:07] hold also for the total wage bill, total inputs, total assets. Uh so this is kind of a generic finding for firm size.  
[01:54:14] These firms are losing market share imports uh and they're uh uh they're essentially contracting their scope. And so how do these firms respond to this negative shock in terms of their management? What I find is that the  
[01:54:28] share of family in top management declines by about seven or eight percentage points which is a 15% decline relative to the pre-policy control group mean. So this is only family firms here.  
[01:54:38] Those firms that have this margin to adjust and to professionalize according to my definition and I find that uh firms start professionalizing after import liberalization.  
[01:54:48] And what's interest uh this is for the firm or the product.  
[01:54:54] I guess I have maybe this image is wrong. I would I would have thought these are very diversified companies and so uh >> so what I'm going to do is the treatment is going to be at the firm level and the way I'm going to go from the product to  
[01:55:07] the firm level is that the I'm going to identify the mean or the highest revenue product for each firm and then just do that. I >> I guess what I was when you have a 60% drop in revenue for the firm I I guess I wouldn't have thought that the top product was 60% of revenue. So that's  
[01:55:21] what I'm that's what I'm I'm missing something. Uh no, so this is total revenue including those. So this is for that firm this is total revenue.  
[01:55:30] >> Yeah. So I I mean the what what's the fact about for the largest product for a firm what percentage of revenue is that typically?  
[01:55:37] >> Okay, I could I could uh I could >> like I would have thought it was much less than 60%. So >> okay I could look at that. Got you.  
[01:55:45] >> And for the managerial response what's interesting is that almost all of this is being driven by baseline lagard firms. So what what I do here is to in the third panel is to split the purple uh middle panel into two parts. In red  
[01:55:58] are firms that are in the lowest tile of pre-policy quantity productivity or TFPQ and in blue is everyone else. And what you see here is that the response for these least productive firms uh it's much larger and there's almost no  
[01:56:12] response for firms uh you know who are uh uh uh who are not less productive who are sort of not in the lowest tile of pre-policy productivity and this is to some extent very different from a story where the frontier firms often in in  
[01:56:26] trade models are adopting better technology or innovating here the action is actually coming from the lagards in the in the economy. Yeah.  
[01:56:35] Just looking at these first two graphs here, it seems like um I would have expected given your story that there would be some recovery of revenue in the outy years once firms respond by changing the you know the managerial.  
[01:56:48] >> No, that's a good point. So now of course you you might say well maybe the green dots would have been even lower if the this reforms hadn't happened but what are your thoughts about this pattern?  
[01:56:56] >> No, that's a good point. I I mean I have some sort of a figure to show you for that. on the left is all firms, family firms, professional firms, right? Only half of them, which is family firms, are doing the professionalizing. So I can split the left figure into two. And  
[01:57:10] actually I find that family firms see a lower decline in revenue relative to professional firms where I don't see any management churn. So that's one of the buttons. Um we can talk about that more, but that's that's something that I can see for example uh in revenue data when I split it by family versus  
[01:57:25] professional. Okay. Yeah. by family and professional firms revenue impacts and then also there may be some churn in the professional firms as well.  
[01:57:35] We need to I guess part of the story the contract are you going to show?  
[01:57:40] >> So I'm not going to show the lack of churn in professional firms. I don't see any churn in professional firms. Um I do have a button for um for splitting revenues by family and professionals.  
[01:57:51] And that that's exactly what I was telling Dean that family firm revenues actually fall by less than professional firms. And that to some extent would be because the family firms are the one that are indogenously responding to the trade shock by upgrading their  
[01:58:04] management by getting fresh talent. >> Yes. And and that was that was contrary to >> Okay. Okay. Okay. Let me show that.  
[01:58:17] Okay. So here you see uh in orange are professional firms and in purple are family firms. Um and you see that the declines are a little bit more for uh professional firms than family firms. My interpretation of this is that only one  
[01:58:31] of these is endogenously responding to their management. It's the family firms who are inside their PPF and now going closer to their PPF because they think they can get outside talent and and make an improvement.  
[01:58:51] I could so maybe one thing I could do is to weight this by >> uh yeah weight this but but also uh you know maybe doing a binary treatment I could have an intensity of treatment um based on how much of your product scope  
[01:59:05] is impacted by this by this shock but yeah maybe I could do a few things quickly.  
[01:59:23] Yeah. Okay. Okay.  
[01:59:29] Great. Um, and so here I showed you that the share of family is going down and that's actually a one is toone replacement of family managers with outside professionals. So on the left here I've got the number of family top managers with declines by about 20% and on the  
[01:59:43] right is the number of non-family uh top managers that's increasing by exactly uh the same magnitude. So there's a replacement going on uh here.  
[01:59:53] Um I also find within family firms within the unproductive family firms that actually managers coming from the same location I've got the city of birth and I can infer the state of birth that that declines as well. Um just to show that you know networks could not  
[02:00:07] necessarily only be family they could be geography to some extent as well and there's some churn going on where you're moving uh you know to people outside the family to people outside your geographic um uh network.  
[02:00:20] And finally on showing management churn here is family firms and professional firms and there's some baseline churn going on um uh uh you know in both of these firms but if you look at the age distribution of who's resigning in  
[02:00:34] family firms versus who's resigning in professional firms uh you actually see this biodality where in the family firms it's either the really young uh you know maybe the idiot son who's now being booted out or the really old 80-year-old  
[02:00:47] sort of patriarch who's now finally pushed out of the family firm but it's more unimodal for uh for professional firms and so the age distribution the variance of the age distribution is declining within family firms. Okay. So that was the second set of results where  
[02:01:02] I showed how management is responding. And now let me show you um two cuts of the data to understand what might be the possible impact of this on firm productivity. And the first is that I'm going to look at baseline professional firms. So firm that have no family  
[02:01:16] before the shock. And you actually see no TFPQ gains for these firms. But if you look at firms uh that are family before the shock, you actually see some large TFPQ gains after the shock. And  
[02:01:29] then what I can do is to understand where to some extent where these gains are coming from. That's a difficult problem in this context. I split this purple line into two where I look at family firms that are professionalizing versus family firms that are not professionalizing. So in red you see  
[02:01:44] large TFPQ gains for those family firms who do professionalize their management which is of course an endogenous variable that I'm splitting on and so you know we have to be careful about so take this as a descriptive statistic of what's happening we have to be careful  
[02:01:58] about how we interpret this um there might be selection involved in who is professionalizing and who is not and I think the model speaks to that more directly but also a firm might be doing many other things not just professionalizing their management so  
[02:02:10] think of this as a catchall gain in um you know uh that's correlated with changing your management. It could be change in product scope. Some of that I do find in the paper or change in plant location or things other than that that  
[02:02:22] I don't observe in my data. Yeah. Sorry.  
[02:02:30] >> Yeah. I I just want to push you a little bit on this timing. I guess like I want the story makes sense and but it would have felt much more gradual. you know that when you replace and with professional management they start to hire younger people they you know do off  
[02:02:44] the checklist of management practices so you're seeing this like immediate jump in the FPQ like tell me a narrative >> so I think um if you look at where these managers are coming from I've said nothing about the supply side here >> a lot of them are actually internal  
[02:02:58] promotions um and so to some extent maybe they know what's wrong with the firm they've been in the firm for 20 years um I haven't done this but I can actually look at the tenure of these How long have they stuck with that firm?  
[02:03:09] That could be one. Um um I also see lots of lots of foreigners actually coming at border direct. If you go to Gorga now for example, there are a fancy apartment buildings where you have a lot of people from China, South Korea coming with that management tal talent. Doesn't speak to  
[02:03:23] a lot about about the timing but uh you know maybe they know what the issues are and and you know but I don't have more than that in this present data. One thing I could do is I have access to the balance um the minutes of the meetings  
[02:03:36] of some of these firms from 1996 onwards. And so that could be interesting to parse uh with AI and look at what kind of changes are happening around the same time as management is changing. What is it that they're doing?  
[02:03:49] Um that could be that could be one thing >> where the managers are coming from. It's interesting you do see a little bit of a dip in the professional firms and if they're geographically concentrated which a lot of industries in India are  
[02:04:04] you could imagine poaching from the other firms your new professional managers do you see evidence of that?  
[02:04:09] >> Uh I think something like 15 to 20% are coming from other firms. Most of them are internal promotions and for a lot of them they're just not in my system. So my data only knows you if you're a board of director in some firm. Uh to some  
[02:04:23] extent you could be poached as someone who wasn't a board member uh to this firm. So, so I don't have visibility on that. Uh, but I see a lot of foreigners coming in who are definitely from outside the system. I see a lot of internal promotions who were people already on the board but now get an  
[02:04:37] executive seat uh to some extent. Yeah, >> if I could just offer a friendly amendment to your prior answer like Sema and I have both been asking about the timing of different effects and you know why would the TFP be rising immediately if the management changes are happening  
[02:04:51] with a lag. I guess like the you know these management changes are certainly just a subset of a very multifaceted set of responses to the uh the import competition shock. So um there's almost certainly other things that the firms are doing that you're not that aren't  
[02:05:06] really you're not really exploring because you don't have the data perhaps.  
[02:05:09] Uh and so you know these TFPQ uh improvements are going to be happening because of all of these different responses and not just the managerial changes. So you don't there doesn't have to be a tight map between the timing of the managerial changes and the timing of various improvements.  
[02:05:23] >> That that's well taken and in fact one of the things that is happening almost instantaneously is that the product scope of the firm is changing and the firms are dropping some of the products which perhaps are difficult to produce now or facing too much competition or are lossmaking um uh and and you know so  
[02:05:38] that could contribute to some of this. Okay, great. Uh so now I'm going to sketch out um a simple model where I'm going to start with monopolistic competition uh with CES demand. Uh firms have heterogeneous productivity. They are born with some productivity that's  
[02:05:52] drawn from a potato distribution. And the payoff of the firm depends on its management choice. And so if it stays family, it's going to get some profits that are a function of its productivity Z. But they're also going to get some non-puny private benefits that I'm going  
[02:06:07] to denote as B. So think of this as a money metric value. So how many dollars is this utility amenity value worth?  
[02:06:14] That's B. If you professionalize your management, the only cost that I have in the baseline model is that you lose your private benefits, but now your productivity gets amplified by some parameter gamma that's greater than one.  
[02:06:27] Okay. Um so what are these private benefits? Um when I spoke to firms it uh it was almost a natural thing for them to say as when I asked them why you organize this way. This is the satisfaction you get from uh organizing  
[02:06:40] the firm around the family. You want your siblings, your daughter, your son to be around you when you're when you're running the firm. Um and this has a long tradition uh in the corporate finance literature. I'm going to make two assumptions that that are going to be crucial for a couple of diagrams that  
[02:06:54] I'm going to show you uh next. The first is that professionalizing management is an absorbing state. So once you professionalize, you can't go back to being a family firm because of reputation costs primarily. So think about a firm that's either listed on the stock exchange or a firm who's dealing  
[02:07:08] with creditors. Once you professionalize, it's not a good look for you to go back to family just because the environment has changed. And so I'm going to treat that as an absorbing state. And second, and this is absolutely crucial for a discontinuity I'm going to exploit around the exit  
[02:07:23] threshold, is that firms are going to be hand-to-mouth. So they can't survive without without positive monetary profits. That's your pi. It's not pi plus b that makes you alive or not. It's really that your monetary profits have  
[02:07:37] to be greater than zero. Okay. And this setup leads to three okay sorry before that um there were a couple of questions on contracting frictions and there were that there's actually a couple of ways one could think about contracting  
[02:07:50] frictions in this setup. Uh the first is just a reduced form wedge on profits in the state of the world where you professionalize your management. So that's the tow you lose to share of your profits because this external manager runs away with them. Or you could think of that as a fixed cost. You could also  
[02:08:04] think of the fixed cost as a fixed search cost for searching for professional managers. These two forces on their own can't explain either the qualitative fact that you have more professionalizing in an environment where my profits are going down in a bad  
[02:08:17] shock. nor can they tell me um you know the selection patterns that it's only the least productive firms. So for now I'm uh switching those off but in the paper I discuss you know the direction of the misspecification that that uh the  
[02:08:29] simplification can cause. Yeah, >> they're risky.  
[02:08:35] >> So, yeah, >> you might increase expected but have the risk.  
[02:08:40] >> Oh, so you mean tow is risky. >> Yes, exact.  
[02:08:48] >> Yeah. So I think one way which is not exactly what you're saying but there's this literature and macro on idea diffusion where what you do is when you adopt a better technology which is like professionalizing your management you pick a new draw from uh a new from from  
[02:09:02] the same paro distribution and I toyed with that which is similar in the sense that once I professionalize I don't really know what my returns are going to be um that model will give me negative selection into which firms are professionalizing but it's actually  
[02:09:16] going to give me a counterfactual cross-section in the pre-policy period because the family firms are never who are very productive will never professionalize because they only have down to go and it's the least productive firms uh and so it's going to give me something that's counterfactual to the data that I showed you where family  
[02:09:31] firms are smaller and professional firms are larger okay let's discuss that offline okay um and so in the baseline model I'm going to switch these uh uh the tow and the fixed cost I'm just going to have  
[02:09:44] this simple uh model here and this leads to three endogenous thresholds and I'm going to show this through the help of these uh these mallets diagrams where there's baseline productivity on the x x axis. So this is the productivity you're born with and on the y-axis are profits  
[02:09:59] and payoffs and so profits are linear in in log productivity or this transformation of productivity and you can see in red here are profits for family firms. There's some threshold below which they're going to make negative profits but these guys  
[02:10:13] don't only care about these monetary profits. they actually care about the profits plus the private benefits.  
[02:10:17] That's the golden curve there. What's crucial is I exit the market when my ne profits are negative. And so my um and so my um private benefit curve is  
[02:10:31] actually going to end here. It's not going to extend all the way here. I'm going to lose my private benefits if I exit and I need positive monetary profits in order to be able to survive.  
[02:10:41] The firm can also choose to professionalize its management in which case its productivity is going to get amplified by gamma. That's going to shift it up and this is going to split the productivity region into these four  
[02:10:54] regions where initially you've got a region where no one is productive enough to survive. So whether your productivity is high or low, you're just going to have to exit. And then second is a region where family firms can't on their own survive unless they professionalize.  
[02:11:08] their profits are negative under family management and so they have to jump up to this blue curve to be able to survive and so this is a region where lagard firms are professionalizing in order to avoid exit and I'm going to argue that trade makes this region larger and leads  
[02:11:22] to more professionalization yeah >> I always thought the tradeoff was just the you know higher productivity the gamma versus the tax of having to pay the outside  
[02:11:37] >> like that's going to so that >> so give me one second and I'll come come to that that's the second threshold the ZD that you're thinking about let me come to that in a minute okay um and so here's a region where lagard firms  
[02:11:51] professionalize to avoid exit and then there's this white region where family firms live because even though their monetary uh even though their monetary profits are less than they would have been if they professionalized their total payoffs are greater than the  
[02:12:06] monetary gains from professionaliz izing and so they stay family here and then finally you have another region why where professionalization occurs uh due to the logic that you're talking about Ben where the frontier firms are going to professionalize because they get  
[02:12:20] higher gains and that gain scales with firm size uh and um um uh because the gains are scaling with firm size and the private benefits are zero you're going to get this positive selection which is similar to many trade models where the  
[02:12:34] frontier firms like in Paula's a job market paper you're going adopt better technology because you know the gains suddenly become with export opportunities become uh become higher.  
[02:12:46] I have a question. I had a thought.  
[02:13:02] >> Yeah. >> And then there's one version of different version.  
[02:13:26] >> Okay. >> Yeah.  
[02:13:32] >> Yeah. >> Sure.  
[02:13:36] >> What's the >> So, I think what I'm looking at is to try and get a sense of when I said in the introduction that there's opposing forces going on here. It's good that firms are professionalizing. That's good for productivity. But it's bad that there's weaker selection. We need to  
[02:13:48] aggregate things up in a way that imposes the minimum assumptions required to get me also predictions that are in conssonance with the event studies that I've showed you. So that's the goal to  
[02:14:01] really put numbers on the aggregation exercise to try and understand that if you have an economy like this, unlike an economy like mellet or a standard trade model where this X and efficiency gain is really taken away, firms are always on the frontier. What's the difference when we think about aggregate  
[02:14:14] productivity or aggregate welfare? >> Oh, just that uh if there was a world  
[02:14:29] where the gains from professionalizing were uncertain. Um what that would do is in the pre-polic I mean in the in any steady state in the economy that would mean that the f that that the family  
[02:14:42] firms are more larger than the professional firms which is counterfactual.  
[02:14:53] Yes. So I'm going to come to that in a couple of slides.  
[02:14:56] >> Sure. Yeah. Okay. So I I I can show you how I compare that to the model and then we can maybe discuss that. Okay. Um and so there are these regions everyone exits the lagards professionalize.  
[02:15:07] There's a region in the middle for family firms and then frontier firms are professionalizing. They only have a monetary trade-off. They're just caring comparing is my B large enough to outweigh the gains or not. Um and uh those guys are at the right side of the  
[02:15:21] productivity distribution. So if you overlay the productivity distribution drawn from a parto this is just this is not the model. This is just um expositional. You'll see that you get professional firms on the left that are going to eventually professionalize um  
[02:15:35] and then family firms in the middle and finally professional firms on the right.  
[02:15:39] So you see this non-monotonicity. But note that this is only in terms of baseline productivity. the professional firms are going to realize these gamma gains and their productivity is going to shift to the right and in the steady state in terms of realized productivity  
[02:15:52] which is the productivity I observe in my data uh it's actually going to match uh quite well the kernel densities that I showed you where family firms were smaller than professional firms okay great um so I want to quickly go over  
[02:16:07] what trade does to a picture like this so here you've got um just a watered down version of the diagram I'm going to model import liberalization as a reduction in the market demand. And so that's going to reduce my profits. It's going to reduce all the curves. Uh it's  
[02:16:21] going to shift them down and it's going to increase and open up an area here where there's going to be trade induced exit because a class of family firms are going to now find that they're making negative profits, but they can give up these private benefits and professionalize their management. So  
[02:16:36] Ben, it's this area which you wouldn't get in a world where private benefits was zero and you only had this trade-off between private uh I mean you know you only had I mean this is why private benefits need to be non-monetary. If they were monetary you wouldn't have this discontinuity around the exit  
[02:16:51] threshold. >> You're just imposing why why couldn't I just >> take a few losses to keep going with the benefits.  
[02:17:03] So I'm imposing it through this idea that firms are hand-to-mouth. And I mean it the assumption boils down to whether banks should keep lending to me if I'm making negative losses. This is a static model. These profits are think of them as net present value. If I'm  
[02:17:17] continuously making um losses, you know, you could give a wealth distribution and make this uh more complicated. There would still be some firms who might not have enough wealth or some firms who might be particularly credit constrained who will have to exit. Uh so think of  
[02:17:32] this. I mean this is a very simple exposition of that idea that these firms can't afford having negative profits. If you can imagine a more realistic world where they have wealth >> problems in dealing with your >> Yeah.  
[02:17:46] >> But it could be that those problems are also changing smoothly. I'm just wondering what role the risk continuity.  
[02:17:54] >> Um no no I think I'm fine with a smooth change. That's fine. It'll still be concentrated in the part of the productivity distribution which is less productivity firms. If it's smooth that's actually fine and that will hopefully actually make me fit the data better. I'm I'm imposing this just for  
[02:18:08] simplicity. Okay. Okay. Um there's also going to be a decline in profits by professional firms. Um uh and there's going to be some exit and regretting of past delegation, but I'm going to uh skip this and come to the quantification  
[02:18:22] uh just uh because of time. Okay. So I'm not going to go into the details of the entire uh calibration pro procedure. I'm going to focus on two parameters the B and the gamma that are the private benefits and the gains from  
[02:18:35] professionalizing which I'm going to fit uh and I'm going to fit the model uh to India's pre-policy economy. So I'm not going to deliberately use the event studies to inform my quantification and that gives me a window to then implement  
[02:18:49] a similar experiment in my model and compare that the responses I see in my model to the event study estimates that I have. Right? So, so that's the deliberate call that I've taken. And the parameters that are uh explaining um the private benefits are the share of family  
[02:19:03] firms in this economy. So, this white area that I showed you where family firms um the total payoff is greater than professionalizing. Sorry, that shouldn't be missing. The estimate is 18. That's about onethird of average profits in the economy. These are pretty  
[02:19:17] large at the firm level. And second are the gains from professionalizing where I'm just going to look at the difference in the log mean revenue between family and professional firms. And that gives me an estimate of about 23% which is a little lower than what I found in the  
[02:19:30] event studies. Um and uh uh but you know it's it's within the within the confidence interval uh band. And then what I can do is to run a policy experiment in the model where I increase  
[02:19:42] import penetration to match what how it changed between 1995 and 2008 the pre-pol uh through the policy period and resolve the model and then compare on a bunch of unargeted moments um how does  
[02:19:55] the model respond in how do the firms in the model respond to what I found in the event studies and so here are the numbers where I've taken the averages of all the post policy uh coefficients of the event studies for revenue declines, profit declines within firm productivity  
[02:20:10] which is actually lower in the model uh than in the data. So my results are to some extent conservative. Uh about 50% of family firms professionalized both in the model and the data. Uh and I have exit rates uh which are within the confidence interval band. So this is the  
[02:20:25] out of sample unargeted moments uh exercise uh that I do with the model.  
[02:20:30] And then finally the purpose of the model was really to try and understand what does all of this mean for aggregate productivity and welfare. And for that I'm going to um uh look at two post policy allocations. The first column is going to be the benchmark economy and  
[02:20:44] the second column is going to be a counterfactual economy where I'm going to shut down this professionalizing uh behavior. So firms are not allowed to professionalize anymore. And what I find is that aggregate productivity uh which is average productivity in the economy  
[02:20:58] increases by about 11.45% um in the baseline economy. And almost all of this is coming from selection. So this idea of within firm professionalizing that I showed you is  
[02:21:10] only uh 83 percentage points about 7.5% of the total productivity gains. And this might seem surprising to begin with because the within firm productivity gains are 30%. But remember that the firms who are professionalizing in the  
[02:21:24] model and the data are only a subset of firms i.e. the family firms and even within that subset it's the least productive firms and so you're only going to get that much action. they're a small part of the economy that are professionalizing and upgrading.  
[02:21:38] Um, if you shut down the channel to professionalize, you actually get larger productivity gains and the larger productivity gains are coming primarily from much greater selection. So now in this counterfactual economy, firms don't  
[02:21:52] have the margin to professionalize their management and stay alive. They're actually exiting uh and that's going to increase um uh productivity uh aggregate productivity.  
[02:22:02] >> Sorry. Okay, let me let me uh so while this is bad for aggregate productivity, it might actually be good for welfare because consumers value variety. And to the extent that you have less variety loss in the benchmark economy, you  
[02:22:15] actually see that when you look at when you compute aggregate welfare, that's actually larger in an economy where professionalizing is allowed because um uh because more varieties are preserved.  
[02:22:26] Okay, so um with that I'm going to uh conclude. I'm uh almost out of time. Uh I talk in this paper about how trade sorry oh I see okay um I talk about how trade uh increases aggregate productivity to  
[02:22:41] some extent by improving lagards uh within firm productivity um and the empirical lens here is family firms but I think this is more general um illustration of an idea that absent competitive pressures you might have um you know you might be able to sustain  
[02:22:56] taste based preferences all becker but competition can maybe move you closer to a meritocratic uh world. Thanks very much.