The Asymmetric Pass-Through of Sovereign Risk

videoType

dynamic-media

videoDmUrl

keyFrameImage

timestamp

00:00 OK.

00:01 So thanks everybody for uh joining in.

00:03 Like,

00:03 uh,

00:04 we have another job market seminar today

00:07 from uh someone who is a familiar face to some of you,

00:11 because Matthias Moretti used to work at the bank before he started his PhD

00:15 and he's going to present to us a very

00:16 interesting paper on the asymmetric posture of sovereign risk.

00:19 So Matthias,

00:20 take it away.

00:21 You have 75 minutes and

00:22 for the 1st 10 minutes,

00:24 we're gonna

00:25 respect

00:25 the no questions rule.

00:27 OK,

00:27 OK.

00:27 Thank you very much,

00:28 Bob,

00:29 for that introduction.

00:30 So this is the title of my job market paper,

00:32 The Assymetic Pathrough of Sovereign Risk,

00:34 and thank you very much for,

00:35 for this opportunity and for attending my presentation.

00:38 So let me jump directly to the motivation of my model.

00:42 And let me start saying that

00:44 it is a very well known empirical fact

00:46 that sovereign debt crises are characterized by very large

00:50 and persistent declines in economic activity.

00:53 So for example,

00:53 think about Italy or Spain.

00:55 It took more than 7 years for their economies to fully recover.

01:00 Now,

01:00 the question clearly here is why is that the case?

01:03 Why are these crises so persistent?

01:06 One of the most common explanations that you will find in the literature

01:09 is based on the exposure of the domestic banks to sovereign risk,

01:14 and the idea in all these studies is that the domestic

01:16 banks typically hold a very large share of government bonds,

01:20 and because of that,

01:22 whenever there is an increase in sovereign risk

01:24 that will weaken banks' balance sheets,

01:27 banks will tighten the supply of credit,

01:29 and that propagates the effects

01:31 to the rest of the real economy.

01:33 Now in my market paper

01:35 what I show is that behind that explanation

01:38 there is a very important missing piece,

01:40 and that is corporate risk.

01:42 So I'm going to show today that sovereign debt crises

01:45 are characterized not only by increases in sovereign risk

01:48 but also by very large increases in corporate risk.

01:52 In the empirical analysis,

01:53 I will address two different questions.

01:55 The first one is to what extent is this increase in corporate risk

02:00 driven by the initial increase in sovereign risk?

02:04 And also I'm going to shed some light on

02:06 the different channels that are driving that particular relationship.

02:11 Now the key issue here is that the domestic banks

02:14 are of course heavily exposed to this corporate risk,

02:18 and because of that,

02:19 this additional increase in corporate risk

02:21 will further affect the balance sheet of the banks and

02:24 banks are going to contract even more the credit supply.

02:28 So to understand that concern or that issue,

02:30 I'm going to build a quantitative model in order

02:33 to study and quantify the role of corporate risk

02:36 as an amplification mechanism of a sovereign debt crisis.

02:42 In terms of empirical analysis,

02:43 the first thing that I'm going to do in

02:45 this paper is to measure the transmission from sovereign risk

02:49 to corporate risk.

02:51 To address that,

02:52 I'm going to use Italy as a case of a study

02:55 and using this heteroschasticity-based approach

02:58 that exploits differences in the volatility of Italian sovereign risk

03:03 around some news events that I will describe.

03:06 What I find is that sovereign risk can account for almost 30%

03:10 of the total increase in corporate risk.

03:12 And on top of that overall effect,

03:14 there are very important heterogeneous effects across the non-financial firms.

03:19 In particular,

03:20 what I find is that the riskier firms,

03:22 meaning those firms with a higher default probability,

03:25 are the ones that are more affected

03:27 by this sovereign shock.

03:30 In the second part of my empirical analysis,

03:33 again using Italian bank level data in this case,

03:36 I'm going to shed some light on the different channels that

03:39 are behind this transmission of sovereign risk to the non-financial firms.

03:43 And although there are many,

03:44 many channels behind it,

03:45 of course,

03:46 what I do in this section is to study and quantify

03:49 the importance or the role of the domestic banks

03:52 in this particular transmission.

03:54 So I'm going to present some evidence that shows that not surprisingly,

03:58 the banks are passing on sovereign risk to their corporate clients.

04:02 In particular,

04:03 the result that I have here shows that

04:05 those banks with higher sovereign exposure before the crisis,

04:09 meaning those banks holding a larger amount of government bonds,

04:13 those are the ones that tend to exhibit.

04:16 A larger increase in their corporate and non-performing loans.

04:22 Now,

04:22 in the second part of the paper,

04:24 and this is my main contribution to the literature,

04:27 I'm going to ask to quantify the importance of corporate risk

04:31 as an amplification mechanism of a soaring debt crisis.

04:34 And the key idea is that there is going to be an interaction between

04:38 the sovereign risk,

04:39 the corporate risk,

04:40 and the financial system as a whole.

04:42 To address that type of method,

04:45 there are 3 key ingredients or elements in my model.

04:49 So first,

04:50 motivated by these empirical results,

04:52 I'm going to feature a model in which firms are heterogeneous

04:56 and in which they can default on their loans.

04:59 The banks,

05:00 the domestic banks,

05:01 are going to provide all the loans to those firms

05:04 and at the same time they're holding government debt,

05:07 and that's why these banks are exposed to both sovereign

05:10 and corporate risk.

05:12 I'm going to assume certain financial frictions that I will describe

05:16 and also this is a model with aggregate uncertainty,

05:19 and the only source of aggregate shock is what I call

05:22 a sovereign risk shock that I will explain in a second.

05:27 Now the key mechanism in the model is that under these three key ingredients,

05:31 corporate risk

05:33 is endogenously linked to sovereign risk through the banking sector,

05:37 and I'm going to use the empirical estimates of the previous analysis

05:40 in order to discipline this particular relationship.

05:44 So the idea here is that

05:45 if there is an increase in sovereign risk,

05:48 the balance sheet of the bank will be affected

05:50 and the bank's lending capacity will be affected.

05:53 So banks will charge higher interest rates,

05:56 and in that context,

05:57 given that firms can optimally choose when to default,

06:00 more firms will have more incentive to default,

06:03 and that increases the default rate,

06:05 but also corporate risk,

06:06 and that extra increase in corporate risk

06:09 will further affect the balance sheet of the bank.

06:12 And because of that,

06:13 The model delivers a two-way feedback loop

06:17 between the corporate risk of these non-financial firms

06:20 and the balance sheet of the domestic banks,

06:23 and I'm going to show you during this presentation

06:25 that this feedback loop

06:27 significantly amplies both the sense and the

06:29 persistence of a sovereign risk shock.

06:35 Finally,

06:36 consistent with the empirical findings,

06:38 the model can capture the asymmetric transmission of sovereign risk

06:42 to the non-financial firms.

06:44 So I'm going to use the model.

06:45 I'm going to exploit this heterogeneity

06:48 in order to provide some policy guidance.

06:53 OK,

06:53 so where does my paper stand

06:55 with respect to this sovereign debt type of literature?

06:59 There are essentially

07:00 3 different categories here.

07:02 So the first category tend to view the fundamentals of the economy as given

07:07 and they study how changes in these

07:09 fundamentals affect the government's incentive to default

07:13 and therefore sovereign risk.

07:16 Now a second strand takes the opposite approach.

07:19 They tend to view sovereign risk as an exogenous shock,

07:22 and they study the channels through which changes in sovereign risk

07:26 affect the real economy.

07:28 In terms of empirical analysis,

07:30 the closest paper to mine is this one by Heber and Schreger for the Argentine case.

07:35 In terms of the quantitative model,

07:37 my paper is mostly related to this paper by Bocola 2016.

07:41 Now there is a 3rd strand of course that I study the feedback effects between

07:46 the real economy or the non-financial firms

07:48 and the sovereign risk.

07:50 For instance,

07:50 this paper,

07:51 new paper Barraano Bay and Bocola.

07:54 So where

07:55 can I classify my papers in these 3 different categories?

07:59 Essentially in the second one.

08:00 So what I'm going to do

08:02 is to view sovereign risk as an exogenous shock.

08:05 I'm going,

08:05 I'm going to contribute to this literature because I'm

08:08 going to explicitly model the behavior of corporate risk

08:11 during a sovereign debt crisis,

08:13 and I'm going to study and quantify

08:16 how the presence of corporate risk

08:18 through the interactions with the financial system

08:20 can further amplify the effects of a sovereign risk shock.

08:26 OK.

08:26 Let me pause here and please let me know if you have any questions.

08:29 If not,

08:29 I will move to the empirical analysis.

08:35 Claudia has a question.

08:36 Claudia,

08:37 I want you go ahead.

08:39 Hi,

08:39 Maths.

08:40 Uh,

08:40 so I have a question here.

08:41 So,

08:42 uh,

08:43 so you're saying that banks with more sovereign debt,

08:46 uh,

08:48 is,

08:48 are gonna tighten their margins to non-financial firms.

08:52 Uh,

08:53 so I'm thinking on,

08:54 uh,

08:55 my work on Mexico.

08:56 So in Mexico,

08:57 I see

08:58 that usually banks that lend more to the government also

09:02 are more likely to lend to firms,

09:04 non-financial firms that are more connected.

09:06 To the government,

09:06 so suppliers of the government.

09:09 So then,

09:10 could this tightening that you observe or how would this enter in your model,

09:14 that this tightening is not coming,

09:16 you know,

09:16 from banks tightening credit supply to a random firm,

09:20 but to,

09:21 to firms that were also exposed to,

09:23 to this uh sovereign shock of the government because they are

09:26 suppliers of buyers or buyers of the government.

09:30 Yeah,

09:30 thank you,

09:31 Claudia.

09:31 That is an excellent question.

09:32 So you will see that through the lens of my model,

09:36 all there is going to be one representative bank

09:38 and the supply will be affected because the

09:41 balance sheet of the bank will be affected.

09:43 But in that sense,

09:44 my model will not be able to capture,

09:47 for example,

09:47 firms that rely more or less

09:49 on public subsidies in order to operate.

09:52 So that will be completely outside the scope of my paper.

09:55 Now in terms of the literature that is already there,

09:57 at least for the Italian case,

09:58 there is a Paper by Botero Lenzu and co-authors in which

10:02 they show

10:03 that the same firm that is operating with two different banks,

10:07 the type of bank that is highly exposed to this sovereign risk,

10:12 contracts the supply even more

10:14 relative to the other bank to this particular firm.

10:17 So even after you control for all these firm level characteristics,

10:20 the type of bank lending channel that I have considered in the model

10:23 is already there in the literature,

10:25 and there are a lot of papers

10:26 that document that type of relationship.

10:31 Alvaro

10:33 Please go ahead and ask your question.

10:34 Uh,

10:35 yeah,

10:35 thank you,

10:35 uh,

10:36 about,

10:36 uh,

10:36 so Matthias,

10:36 one quick,

10:37 quick question.

10:38 So

10:38 I mean,

10:39 naturally when there's,

10:40 uh,

10:40 sovereign risk or where when

10:43 trade agencies downgrade the sovereign,

10:45 uh,

10:46 sovereign,

10:47 uh,

10:47 all or the private is somewhat,

10:49 you know,

10:50 benchmark against the,

10:51 the

10:52 government rating.

10:54 So

10:54 naturally when

10:56 the rating of the government goes is lower than.

10:59 All firms in the country will be downgraded

11:01 or this the direct effect of,

11:04 of,

11:04 you know,

11:05 more costly is that

11:07 that what you have in mind here or not necessarily because you're,

11:11 yeah,

11:11 thank you,

11:11 Alvaro.

11:12 Uh,

11:12 this is not exactly the type of mechanism that I have in mind.

11:16 I know that there are a couple of papers that

11:18 exactly exploit that type of of of of credit ratings

11:22 between the corporate and the sovereign government

11:25 because there is kind of like a ceiling

11:27 in which the corporates cannot be graded above

11:29 the sovereign.

11:30 That it will not be present at least in my empirical or quantitative model,

11:34 but I do agree that it's another channel through which

11:37 the changes in sovereign risk may end up affecting

11:40 the corporate risk,

11:41 and that is a channel that works completely outside the bank lending channel.

11:45 And I agree with you that may be present in the effects

11:47 that I'm going to capture today,

11:49 but that is not what I'm going to use in terms of the identification of the effects.

11:52 I'm going to rely,

11:53 as I'm going to explain it on some news that are coming from other countries.

12:06 OK.

12:07 OK,

12:07 so

12:09 I think you were as well.

12:12 Oh sorry,

12:13 Sergio,

12:13 go ahead and ask your question.

12:16 Um,

12:17 thank you,

12:18 Mathias.

12:18 One question,

12:19 you place your,

12:20 your paper more on the second,

12:22 uh,

12:23 strand of the literature,

12:24 but by increasing corporate risk and affecting the real economy,

12:28 you can also hit,

12:29 uh,

12:30 the sovereign risk.

12:31 Easily.

12:32 Uh,

12:32 I,

12:33 I understand that you might not capture that in your model,

12:35 but that's,

12:35 that would be something that will affect the sovereign as well.

12:39 Yes,

12:39 uh,

12:40 as,

12:40 as you said,

12:40 in my model,

12:41 I'm not going to consider purely for simplicity,

12:44 of course,

12:44 the feedback effects between the real economy

12:47 and the sovereign risk of the government.

12:49 But you will see that I'm going to calibrate the model in order

12:52 to target the causal effect going from sovereign risk and corporate risk.

12:57 So I'm going to estimate that on the empirical analysis,

12:59 and I'm going to use that in order to calibrate the model

13:02 because of course my model won't be able to say anything at all

13:05 regarding this feedback loop between

13:07 the fundamentals of the economy and sovereign risk.

13:13 Art

13:16 Um,

13:16 hi,

13:16 Matthias.

13:17 Uh,

13:18 uh,

13:18 uh,

13:18 you can postpone this to the empirics if you,

13:20 if you prefer,

13:21 but,

13:22 um,

13:22 I'm just wondering how you're going to,

13:24 uh,

13:25 distinguish,

13:26 um,

13:26 spillovers from,

13:28 from

13:28 sovereign risk to corporate risk from,

13:31 you know,

13:31 very standard kind of common shocks that are going to be going on.

13:34 Like,

13:34 you know,

13:36 uh,

13:36 depreciation of the exchange rate is going to be correlated with sovereign risk and

13:40 have a standard balance sheet channel effect on,

13:42 uh,

13:42 on,

13:42 uh,

13:44 Yes,

13:45 yes,

13:46 let,

13:46 let me postpone that discussion because that is the main identification concern,

13:50 and I will be,

13:51 I will come back to that point because it's one

13:53 of the most important points in the empirical identification strategy.

13:59 Fun,

13:59 go ahead.

14:01 Yeah,

14:01 one clarification,

14:02 so you're,

14:03 you're considering

14:04 firms in the country for which government bonds have been downgraded?

14:09 Is that the,

14:10 is,

14:10 is it going to be a,

14:11 a,

14:11 a country-level analysis

14:13 or do we consider also all the countries that

14:16 own

14:18 sovereign bonds,

14:19 on the bonds of a given country for which it's downgraded?

14:22 Uh,

14:23 this is going to be a

14:24 country-level type of analysis because I'm going to focus

14:27 on Italy only as a case of a study.

14:30 Uh,

14:30 and the idea is that I'm going to use some news

14:33 coming from other countries in order to identify the effects,

14:36 but at the end of the day,

14:37 this is

14:37 a result

14:39 that is based on the Italian economy.

14:43 Yeah,

14:43 so my question,

14:44 for example,

14:44 if a lot of countries own US debt,

14:47 yes,

14:47 that doesn't apply,

14:48 that would not apply to the US in a sense.

14:51 It has to be the case that a lot of the debt is owned by

14:54 domestic,

14:55 exactly.

14:55 And,

14:55 and in Europe as a whole,

14:57 the domestic banks in Italy are holding a significant share

15:01 of the Italian government bonds.

15:03 Of course in the US that is,

15:05 that is different for sure,

15:06 yes.

15:07 But this is not specifically something regarding Italy because it's something

15:11 that you usually tend to observe in other European countries,

15:14 Spain,

15:15 Greece,

15:16 Ireland,

15:16 and so forth.

15:17 Uh,

15:17 now,

15:18 I,

15:18 I don't want to delay you further,

15:20 but

15:21 Will you be discussing in the,

15:23 in this context also

15:26 the role of

15:27 regulation or moral suasion because in Italian banks,

15:32 as you know,

15:33 you know,

15:34 and especially during the height of the Greek crisis,

15:37 the government was Basically lining up the

15:39 banks telling them to participate in auctions.

15:41 So you know,

15:43 there's an element whether you want to call it,

15:45 you know,

15:46 in that particular example,

15:47 moral suasion or

15:49 or regulatory changes that get the banks to buy

15:53 the more debt.

15:55 That,

15:55 I mean that is an excellent question.

15:56 That will be completely outside the scope of my paper.

15:59 I know there are papers,

16:00 for example,

16:00 by Ivain in which they explicitly study this

16:04 concern regarding moral suasion,

16:06 and that is something I'm not going to consider,

16:07 at least in this version of the paper.

16:14 OK,

16:14 great.

16:15 Darren,

16:15 no more questions for now,

16:16 so feel free to continue.

16:18 OK,

16:18 OK,

16:18 thank you.

16:19 So let me jump to the empirical section.

16:22 I will try to be brief here,

16:23 at least to highlight the

16:25 the identification strategy and the main results,

16:27 but I would like to emphasize more

16:29 the second part of the paper,

16:30 which is my quantitative model.

16:32 So I will give you at least

16:34 some highlights or not the details of what I did,

16:37 but just the big picture.

16:38 OK.

16:39 So as I said,

16:40 uh,

16:40 I'm using Italy purely as a case of a study

16:43 during the last European debt crisis,

16:45 and I have two goals in this empirical section.

16:48 The first one is going to estimate this

16:50 effect going from sovereign to corporate risk,

16:53 and that will be a central part of my

16:54 analysis because I'm going to use those estimates later on

16:58 to calibrate the quantitative model that I will be presenting next.

17:02 As a 2nd goal,

17:03 I'm going to highlight at least

17:05 the importance of the domestic banks in this

17:07 transmission of sovereign risk to the non-financial firms.

17:11 Of course,

17:12 in order to address those two goals,

17:14 the first thing that I need to do

17:16 is to construct a measure of corporate risk.

17:19 And you will see that the type of identification strategy that I'm using

17:23 requires high frequency data,

17:25 basically daily data,

17:27 and for that reason,

17:28 I'm going to compute a proxy of corporate risk

17:31 following the distance to default approach by Merton.

17:35 The idea is that I have a panel of 120 publicly traded non-financial Italian firms,

17:41 and this method allows you to combine

17:43 the daily stock price

17:45 together with some annual balance sheet level data in order to compute

17:49 the risk neutral default probabilities of each of the firms that I have in my sample,

17:54 and that's what I'm going to interpret

17:56 as a good proxy of corporate risk.

17:59 Now,

17:59 why I'm using this measure and not other measures such as

18:02 changes in the CDS spread

18:04 or the corporate spreads of the bonds?

18:07 Because by doing this,

18:08 I can have access to a larger pool of non-financial firms.

18:12 And because I want to describe

18:14 this heterogeneity across the non-financial firms,

18:17 having a larger pool of firms is a central element of the analysis.

18:24 OK.

18:25 Let me skip the details of this merchant approach.

18:27 Let me just jump to the results that I have.

18:30 So in this figure,

18:31 you have this measure of corporate risk that

18:33 is in red based on the merchant framework

18:35 and in the black line,

18:36 you have the measure of sovereign risk

18:39 based on the 10 year CDS spread

18:41 of the Italian government.

18:43 Now,

18:43 as you can see,

18:44 it is very clear,

18:45 there is a positive correlation between the two measures

18:48 and as the crisis unfolds between 2010 and 2012,

18:52 there is not only an increase in sovereign risk

18:54 but a very large increase in corporate risk.

18:57 And the key thing that I want to emphasize here is that

19:01 behind these overall trends or these aggregate trends,

19:04 there is

19:05 a very important degree of heterogeneity across the non-financial firms,

19:09 and that is something that you can observe in this picture.

19:12 So here you have the same measure of corporate risk.

19:15 But the black line shows the median level of

19:18 this risk for the panel of firms that I have

19:21 and the gray area shows a measure of the dispersion.

19:24 So this is the inter-quantum range.

19:26 So what you can see is that

19:28 as the crisis develops,

19:30 there is not only an increase in the median level of risk,

19:33 but also a very large increase in the dispersion of this risk.

19:37 So this highlights at least that there is a

19:39 very important heterogeneity across the non-financial firms during this

19:44 sovereign debt crisis.

19:48 OK.

19:49 Going now to the,

19:51 if I may just quickly interrupt,

19:53 uh,

19:54 how

19:55 do we know really it's the sovereign debt crisis,

19:58 right,

19:58 because there's also a business cycle undergoing this,

20:01 and,

20:02 uh,

20:02 you know,

20:03 the spreads between high yield corporates.

20:07 And high grade corporates widens

20:10 during downturns,

20:12 uh,

20:12 economic downturns,

20:13 you know,

20:13 has a big cyclic,

20:15 uh,

20:15 component,

20:16 and here we're just looking,

20:17 you know,

20:18 at,

20:18 at,

20:18 at one part of the cycle.

20:20 No,

20:20 no,

20:20 I,

20:20 I completely agree with you.

20:21 I mean this is just to highlight this asymmetry,

20:24 uh,

20:24 and now with identification strategy using high

20:27 frequency data at the daily frequency,

20:29 I will be able to identify

20:31 that heterogeneity that is coming from sovereign risk.

20:33 This is just to give you the big picture and nothing more than that.

20:40 OK.

20:43 OK,

20:43 so going now to,

20:44 to the identification strategy,

20:46 what I'm going to do is to estimate

20:47 this effect going from sovereign to corporate risk.

20:50 Now there are obvious challenges when you want to do this.

20:53 The first one is related to causality because it may

20:56 be the case that the crisis started in the non-financial sector

21:00 and then propagated to the sovereign risk of the Italian government.

21:04 Another concern,

21:05 as Art was mentioning at the very beginning,

21:07 is regarding common factors.

21:08 So this is a period in which there was a lot of volatility in Europe as a whole.

21:13 So there may have been an increase,

21:14 for example,

21:14 in risk aversion or an increase in the market price of risk

21:18 that is driving at the same time both sovereign and corpo risk.

21:22 In order to address those types of concerns,

21:25 this is the identification strategy that I proposed in this paper.

21:29 So I'm going to rely on a set of what I call 4 news events

21:33 that will allow me to capture in a sense,

21:35 the exogenous variation

21:37 in Italian sovereign risk.

21:39 So these 4 news events are news that are coming from two other European countries,

21:45 Greece and Portugal,

21:46 and I have two different sets of news.

21:49 I have news about

21:50 bailouts for these 2 other economies and

21:52 also news about some credit rating downgrades.

21:56 The idea is that this news will have an effect

21:59 on

22:00 the Italian sovereign risk that is not related

22:03 to the fundamentals of the Italian economy,

22:05 and I will come back to that point.

22:07 So overall,

22:08 I have a set of almost 30 events between the three big years of the crisis,

22:12 2010,

22:13 202011,

22:14 and 2012.

22:16 So the question is,

22:16 OK,

22:17 I have this set of events.

22:18 How am I going to use those events to estimate

22:21 the effect going from sovereign to corporate risk?

22:25 Formally,

22:26 what I'm going to do

22:27 is to use this heteroschasticity-based approach by Rigo Bonanzak.

22:31 And the idea of this approach

22:33 is that I'm going to exploit

22:35 the differences in the volatility of Italian sovereign risk

22:39 around

22:40 these foreign news events.

22:42 So this same identification strategy is the one

22:45 used by,

22:46 for example,

22:46 Heber and Schreger in the context of the Argentine sovereign debt crisis,

22:51 but it's also widely used in the monetary policy type of literature.

22:55 Rigobonazaki is one example.

22:57 Another example is this paper by Nakamura and Steinson.

23:02 So let me give you at least uh a brief summary of what is all this method all about.

23:08 OK.

23:08 And in order to do that,

23:10 let me first describe

23:12 the set of events that I have captured with

23:14 this foreign news coming from Greece and Pug.

23:17 On the graph on the left,

23:19 on this scattered plot,

23:20 what you observe is the change in Italian sovereign risk.

23:23 On the Y axis,

23:24 you have the change in Italian corporate risk.

23:26 The red dots are for all these 29 events that I have.

23:30 The blue dots are for all the other days that I have in the sun.

23:33 It is clear from the picture that there is

23:35 a positive correlation between the two of these measures,

23:39 not only during the event days but also during the non-event dates.

23:44 Now,

23:44 the table that you have on the right

23:46 displays some moments around a 3 day symmetric window around these 4 news events.

23:52 The only thing that I want to highlight from this

23:54 table is this row that is highlighted in blue.

23:58 So notice here that the changes

24:01 in the volatility,

24:02 sorry,

24:02 in changes in Italian sovereign risk

24:05 is higher during the event days compared to the non-event days.

24:10 And you can show that that difference is strongly significant.

24:13 The key idea of this Rigobon and Zack type of approach

24:16 is to exploit

24:18 the differences

24:19 in these volatility.

24:22 So let me

24:23 Now,

24:23 go to the details of this particular framework.

24:27 Consider

24:28 the following very simple system of equations,

24:30 OK?

24:30 In which the change in corporate risk is just a linear function of

24:35 whatever change is happening in Italian sovereign risk,

24:38 some common factors X

24:40 and some shocks,

24:41 OK?

24:42 And the same goes for the change in Italian sovereign risk.

24:45 In this case,

24:46 this is the function of corporate risk,

24:48 some potential of certain common factors,

24:50 and some shocks.

24:53 Now,

24:53 what I'm going to do

24:54 is to classify

24:57 all the days that I have in the sample between the event

25:00 and the non-event days using these four news events.

25:04 And the formal identifying assumption behind this approach

25:07 is that

25:08 the variance of this a shock,

25:11 meaning the shocks to Italian sovereign risk,

25:14 is significantly higher during these foreign news events,

25:17 and there is a whole literature of sovereign contagion

25:20 that shows and provides evidence

25:22 that backs up this type of assumption,

25:24 OK.

25:25 But now,

25:26 the second part of the,

25:27 of the assumption is that

25:29 the variant of this epsilon shock,

25:31 meaning the shocks to Italian corporate risk

25:34 and also the variance of this potential observed common factors remain the same,

25:40 OK?

25:41 Under those assumptions,

25:42 the exclusion restriction can be interpreted as follows.

25:45 The idea is that the foreign news events coming from Greece and Portugal

25:49 are only going to affect the Italian corporate risk

25:52 through their effect on the Italian sovereign race.

25:56 Under these three main assumptions,

25:58 this Rigobon and SA type of approach

26:00 uses these differences in the 2nd order moments in the volatilities

26:05 in order to estimate

26:07 my coefficient of interest,

26:08 which in my case is this alpha 1.

26:11 So in the paper,

26:13 I do provide ample evidence that backs up this type of assumption.

26:16 For example,

26:17 I show that there are no important trade

26:19 and financial links,

26:20 and I also study what was the impact of this news

26:24 on other European countries,

26:26 on the volatility of the European stock market,

26:29 meaning the market price of risk,

26:30 and so on.

26:31 I would like to skip the details of that because I want to emphasize

26:35 the quantitative model that is coming next.

26:38 So if you have any questions,

26:40 I guess that I will be more than happy to talk about after the presentation,

26:44 but if you don't mind,

26:45 I would like to present my main results and

26:47 then move on and discuss my quantitative model.

26:53 OK.

26:55 In terms of the results,

26:57 what you observe here is this alpha 1

27:00 parameter or the estimation of that parameter.

27:03 The first two columns shows both the IV estimate based on

27:06 this Rigobonasack type of methodology and also the OLS estimate.

27:11 The last two columns show the same set of results,

27:13 but after you include some global controls,

27:16 for example,

27:17 changes in S&P,

27:18 changes in the VIX index.

27:22 Now,

27:22 if you,

27:22 you can see that all the estimates are positive and significant,

27:25 of course.

27:26 And if you take this value that is highlighted in blue,

27:29 this is saying that a 1% point increase in the Italian sovereign risk

27:35 leads to

27:36 almost a 30 basis point increase

27:38 in Italian corporal risk.

27:41 In terms of economic magnitude,

27:43 this estimate is accounting for almost 30%

27:46 of the total increase in Italian corporate risk.

27:48 So it's a very large number.

27:51 Now this estimate is in the ballpark of previous empirical analysis,

27:56 and my contribution to this literature is to show that behind this overall effect,

28:02 there are very important heterogeneous effects

28:05 across the non-financial firms,

28:07 and that is the picture

28:09 that you can observe here.

28:10 So here you have the same IV coefficients

28:13 based on this RigobonaA type of methodology.

28:16 But

28:17 for firms with different levels of risk.

28:22 So the first dot and line that you have up here

28:25 is the IV estimate and the confidence interval.

28:28 When I include all the forms that I have in the sample.

28:31 The next two cases are

28:32 those in which I split my sample in 21

28:36 in which I only include the safe forms,

28:38 and the other one in which I only include the risky firms,

28:41 and the distinction there is

28:43 the median level of pre-crisis risk.

28:46 Now to the right of this red line,

28:48 you have the same set of results

28:50 that for the finer decomposition of firms.

28:53 So firms here are sorted from the safest

28:56 all the way to the riskiest,

28:58 based on their pre-crisis levels of risk.

29:01 Now,

29:01 the key takeaway from this picture

29:03 is that those firms

29:05 that were safer before the crisis

29:08 are almost unaffected by the increasing sovereign risk.

29:12 Now,

29:13 once you focus on this subset of riskier firms,

29:16 those firms are significantly more affected whenever there is an increase

29:20 in the default probability of the Italian government.

29:24 So there is a very important heterogeneity across these financial firms,

29:28 these non-financial firms,

29:29 and that will be a central element,

29:31 as you will see

29:32 in the design of my quantitative model,

29:35 and I will use these different reaction to

29:38 increases in sovereign risk in order to provide

29:40 some policy guidance on the optimal way

29:43 to dampen the negative effects of this type of recession.

29:50 OK.

29:52 OK,

29:52 so far,

29:53 I have not said anything at all regarding the channels that

29:57 may be behind this relationship between sovereign risk and corporate risk.

30:02 So what I'm going to do now

30:03 is to highlight at least

30:06 the role of the banks,

30:07 or what I call the bank lending channel

30:09 in this transmission from sovereign to corporate risk.

30:13 So in this very simple exercise,

30:16 what I'm going to do

30:17 is to use Italian bank level data.

30:20 So I have data coming from the Italian Banking Association that covers

30:23 the universe

30:24 of all

30:26 the Italian domestic banks.

30:28 And what I'm going to do is to exploit

30:30 the within region

30:32 banks heterogeneity across sovereign exposures.

30:35 So of course,

30:36 this is not a causal estimate what I'm going to

30:38 present here that is far from being causal estimate.

30:41 It's just to highlight that those banks that are more exposed to sovereign bonds

30:47 will

30:48 affect or will restrict more their credit supply and that will affect more

30:52 the non-financial firms.

30:52 So that is the idea of this particular section.

30:55 Controlling for this

30:57 regional heterogeneity is important in the Italian

30:59 case because the domestic banks in Italy

31:02 are quite different depending on the region.

31:04 So for example,

31:05 banks in the south tend to be smaller,

31:08 tend to be

31:09 more concentrated or more exposed to sovereign risk,

31:11 while banks in the north

31:13 are typically the larger banks with

31:15 a lower exposure

31:17 to this type of risk.

31:19 And what I'm going to do

31:21 is to study or to quantify

31:23 the differential impact on non-performing loans

31:26 for banks with different levels of sovereign exposure.

31:30 Let me skip the details of this particular section.

31:32 Let me just mention the particular result that I have.

31:35 And what I show in the paper is that

31:37 those banks with a higher sovereign exposure,

31:41 after controlling for this regional heterogeneity and a

31:43 lot of bank characteristics such as the size,

31:46 the leverage,

31:47 the net worth,

31:48 the reserves,

31:48 and many more,

31:49 they tend to exhibit

31:51 a larger increase in their corporate non-performing loans,

31:54 which is kind of like a good proxy

31:56 for

31:57 the corporate risk of their clients.

32:00 The result that I have in particular says that

32:02 if you take two banks that are one standard deviation apart

32:06 in terms of their sovereign exposure,

32:08 the bank with the higher exposure

32:10 tends to exhibit an additional 15% point increase

32:14 in the growth rate of its non-performing loans,

32:17 and that estimate

32:19 can account for almost 20%

32:21 of the total increase

32:22 in corporate non-performing loans that we observed during the recession.

32:26 Now of course this is not a causal estimate.

32:28 It is far from doing that,

32:29 but I'm going to use this

32:31 as a motivation for

32:33 the quantitative model that I'm going to explain next.

32:36 The idea is that

32:37 these banks are an important driver

32:39 in the transmission of sovereign risk

32:42 to the non-financial firms.

32:45 OK.

32:46 Please let me know uh if there is a question.

32:48 If not,

32:49 I'm happy to continue with the quantitative model.

32:54 No,

32:54 there,

32:54 there are no questions,

32:55 but I'm gonna abuse my,

32:56 uh,

32:56 power

32:57 of the chair,

32:58 last one.

32:58 So,

32:59 so how are you distinguishing sort of the bank lending channel from,

33:02 for instance,

33:03 the fact that some firms may be more reliant on

33:06 public procurement or government responders spending?

33:08 That's the part I perhaps missed.

33:11 Uh,

33:11 yes,

33:12 yes,

33:12 I mean,

33:13 behind this,

33:14 this result

33:15 that I show you

33:16 right here,

33:17 this is the overall effect.

33:18 This is the total transmission going from sovereign to corporate risk.

33:21 And as I said,

33:22 behind this result there may be a lot of channels.

33:24 The bank lending channel is one,

33:26 the one that you're mentioning is another one.

33:28 A fiscal channel may also be important or a trade channel may also be at play.

33:33 Now the only thing that I'm highlighting here is

33:36 in this very simple exercise that suggests that

33:39 the band lending channel can explain at least

33:41 for a significant part of the total transmission of sovereign risk.

33:45 I'm not saying that this is the only channel that matters.

33:47 In fact,

33:47 there are more,

33:48 but I'm going to take this as a motivation

33:50 for the model that I'm going to present next.

33:54 Understood.

33:54 Thank you.

33:56 Sure.

34:00 OK.

34:01 OK,

34:02 so,

34:04 so sorry,

34:04 yes,

34:04 yes,

34:04 of course.

34:06 So

34:07 we,

34:07 we can think that the,

34:08 that when banks take uh exposures,

34:11 sovereign exposure and who they lend to is not necessarily completely

34:16 exhausted,

34:16 completely independent.

34:17 I can imagine that.

34:20 If I'm,

34:20 uh,

34:21 I can imagine the local government telling,

34:23 OK,

34:23 you're going to buy not only you get a lot of certain banks,

34:26 but my friend here and there,

34:27 you're going to lend them money for their

34:29 pet projects too.

34:31 So,

34:31 so that will create a lot of correlation

34:34 between

34:35 sovereign exposure and corporate.

34:38 Of course,

34:38 that is a,

34:39 that is a fair comment.

34:40 Uh,

34:41 of course,

34:41 I'm not going to control for all that.

34:43 That's why this is far from being a causal estimate,

34:45 but there is

34:46 a whole literature

34:48 that basically analyze the type of question that you have in mind.

34:51 So this literature,

34:52 what they're going to do

34:53 is to use the type of Kawaa M and type of within firm regression.

34:58 So they're going to consider

35:00 one firm

35:01 that is linked to two or more banks.

35:03 And they're going to study the changes in the credit to that firm

35:08 for banks with different levels of exposure.

35:10 So in that way you can perfectly control for firm level characteristics,

35:13 and what they show is that

35:15 those banks with higher exposure

35:17 tend to decrease even more

35:19 their supply of credit to that particular firm.

35:21 So that of course

35:23 it is a causal estimate in that sense and it's controlling for all the concerns

35:27 that you may have between the sorting across banks and

35:31 the clients of that particular bank.

35:34 And that is of course consistent with the result that I'm showing you here today.

35:42 Al,

35:43 please go ahead and ask your question.

35:45 Uh,

35:46 yes,

35:46 Matthias,

35:47 uh,

35:47 just very quickly,

35:48 so I know you were saying you know you see spreads just simply goes,

35:52 I mean,

35:52 naturally,

35:52 those will be more like financial firms that have

35:55 credit default swaps,

35:57 and you want to have as large as a sample as,

35:59 as possible with the publicly traded companies.

36:02 But I mean,

36:02 just to get,

36:03 I,

36:03 I know you don't want to push a lot on the,

36:04 on the methodology of

36:06 the outcome,

36:06 but

36:07 Uh,

36:08 when you have looked at,

36:09 uh,

36:09 like individual like directive,

36:11 um,

36:12 you know,

36:12 the,

36:12 the implicit

36:13 risk pricing for these firms like looking at,

36:16 you know,

36:17 options and

36:18 Because my sense is that you're saying,

36:19 OK,

36:19 safer firms are not affected.

36:21 Uh,

36:22 riskier firms are when there's these 29 events,

36:25 but a lot can be just because of uncertainty,

36:27 and then

36:28 that might be.

36:28 So

36:29 let me tell you what,

36:30 what I did,

36:30 uh,

36:30 to control for all those concerns that is more regarding common factors,

36:34 for example,

36:34 a change in the market price of risk that may affect

36:37 different,

36:38 uh,

36:38 the safe and the,

36:40 the risky firms.

36:41 So what I do as a check is

36:45 to replicate the same analysis that I just did for Italy but for Germany.

36:48 So I'm going to control or compare a firm,

36:51 let's say in Italy with a 3% default probability

36:54 with a German firm with the same default probability.

36:58 And basically they are controlling for these differences in

37:01 the profile of risk across the different firms.

37:03 And what I show in that analysis that I have is that

37:07 These foreign news events that I captured

37:10 based on Greece and Portugal

37:12 are only significantly affecting the corporate risk of the Italian

37:15 firm but not the corporate risk of the German firm.

37:18 So that at least provides evidence that goes in line with the assumption that there,

37:22 that there wasn't a big change in the market price of risk

37:25 across Europe because

37:27 of these foreign news events coming from Greece and Portugal.

37:33 Thank you.

37:35 It's Sergio and after that,

37:37 Carmen.

37:39 Um,

37:39 Maths,

37:40 uh,

37:40 I understand why you're

37:41 analyzing,

37:42 uh,

37:43 listed firms,

37:44 but by analyzing listed firms you're also constraining

37:48 your analysis to safer firms.

37:50 So you're looking at riskier firm within the safe firms.

37:54 Uh,

37:54 if you were to expand this analysis to all Italian firms,

37:57 you might see much larger effects.

38:00 And I was wondering why you haven't analyzed that

38:02 if you have access to the Italian data,

38:05 and I understand that measuring risk for those firms.

38:11 It will be harder,

38:11 but the transmission channel might be much larger.

38:12 I completely agree with you.

38:14 All these estimates,

38:15 because I'm focusing on this subsample of firms that are large and publicly traded,

38:19 can be interpreted as a lower bound in that sense because these are the firms

38:23 that 1 may think that are going to be less affected by this sovereign shock

38:27 because they have

38:28 other sources of financing or they have more capital,

38:31 more capital buffers,

38:31 and so on.

38:33 Now,

38:33 I'm focusing on those farms because I need the daily

38:37 frequency of the data in order to identify the shots.

38:40 So even if I have access to let's say

38:42 the universe of Italian firms based basically a small and

38:47 medium enterprises,

38:49 even in that case I won't be able to

38:51 use this type of identification in order to estimate

38:54 the pass through going from sovereign to the corporate sector.

38:57 The type of identification

38:59 would be different in that case,

39:00 and the one that I'm using certainly

39:02 won't,

39:02 won't do the trick there.

39:08 Carmen

39:10 Uh,

39:11 yes.

39:11 Um,

39:12 now,

39:13 if you're a bank and you care about the overall risk of your,

39:19 not just loan portfolio but your overall portfolio,

39:22 for better or worse,

39:24 you know,

39:24 uh,

39:25 we do view

39:26 bonds often as the risk-free.

39:29 Uh,

39:30 uh,

39:31 assets.

39:32 So,

39:32 you know,

39:33 ex ante,

39:35 uh,

39:36 is it not plausible that the banks

39:38 that have a higher share of,

39:41 have more exposure to,

39:43 to sovereigns.

39:45 Uh,

39:45 were willing to make riskier loans,

39:48 uh,

39:49 in the first place.

39:50 Uh,

39:52 so,

39:52 you know,

39:53 and when

39:55 the sovereign risk was going on and this is going up,

39:57 this is also the period in which

40:00 economic activity,

40:01 you know,

40:02 was,

40:02 was contracting.

40:03 You,

40:03 you,

40:04 you know,

40:04 you were in recession.

40:05 So

40:06 indeed,

40:06 you know,

40:07 you'd see higher

40:09 NPLs,

40:11 uh,

40:11 but,

40:11 but,

40:11 but that,

40:12 that.

40:13 The overall portfolio risk

40:16 decision.

40:17 Yes.

40:17 Uh,

40:18 you know,

40:19 what was,

40:20 I do have,

40:20 I do have an answer for that,

40:21 Carmen.

40:22 Thank you.

40:23 Uh,

40:23 of course that is something that I won't be able

40:25 to show because I don't have that type of data.

40:27 Now,

40:28 the literature shows that this

40:31 different sorting in the portfolio of the bank or these firm level characteristics

40:36 are not driving the results.

40:38 So this literature that I was saying.

40:39 To Alvaro before in which

40:41 they perfectly control for firm level characteristics using the

40:45 same firm that is linked with two different banks.

40:48 That literature shows

40:50 that the firm level characteristics of the firm are not

40:53 necessarily driven the result that they document in that sense.

40:57 So even though

40:58 this result cannot say anything regarding that because

41:01 I don't observe that type of data,

41:02 I don't have that

41:03 granularity in the,

41:04 in the data set that I'm using,

41:06 I'm confident that these results are not driven

41:09 by that sorting that you,

41:11 that you are thinking because there are many papers

41:14 that show that that sorting

41:15 was not really problematic,

41:16 at least

41:17 in the Italian case.

41:25 Please continue.

41:26 OK.

41:26 OK,

41:27 thank you.

41:28 OK.

41:28 Now,

41:29 let me jump directly to my coin identity model because

41:33 as I said at the very beginning,

41:34 this is the main contribution of my paper

41:36 to this sovereign that type of literature.

41:39 This is a brief picture of the model that I'm going to consider.

41:42 So we have 4 different sectors,

41:44 of course,

41:44 the government,

41:45 the domestic banks,

41:46 the households,

41:47 we are just going to be savers,

41:49 and the corporate sector which is formed by heterogeneous firms.

41:54 Now you can think of this as a closed economy,

41:56 OK?

41:56 And the key ingredients or elements are the following.

41:59 So first of all,

42:00 this is going to be a model in which both the government debt

42:04 and also the corporate sector debt is risky because

42:07 they can default.

42:08 In that sense,

42:09 this is a model with limited commitment.

42:11 Now,

42:12 as I said at the beginning,

42:13 I'm going to assume for simplicity that sovereign risk is completely exogenous,

42:17 so I'm not going to consider feedback effects between the real economy

42:21 and the riskiness of the government.

42:23 In this model,

42:24 banks are going to supply the funds both to the government.

42:28 They are holding the government bonds

42:30 and also they're going to provide loans to the non-financial firms.

42:34 And I'm going to consider some financial frictions,

42:36 in particular,

42:37 a leverage constraint.

42:39 And because of that leverage constraint in which banks are going to be subject to,

42:44 the model will feature

42:46 the thing that corporate risk is endogenously linked to sovereign risk

42:51 through the banking sector.

42:52 And this is the idea.

42:54 There are 2 different mechanisms in this model

42:57 through which sovereign risk

42:59 will end up affecting the non-financial firms.

43:01 I do have an endogenous channel which I call the bank lending channel,

43:05 and I also have an exogenous type of mechanism.

43:08 Now the indulgence mechanism is the following.

43:10 If there is an increase in sovereign risk,

43:13 because banks of course are heavily exposed to that type of risk,

43:16 the net worth of the bank will decline,

43:18 and because of the leverage constraint,

43:20 banks' lending capacity will be affected.

43:23 Banks will reduce

43:25 the amount of loans to the private sector,

43:27 or they will charge higher rates.

43:29 In that case,

43:30 the firms have more incentive to default because now

43:33 the interest rate is higher.

43:35 They find it more difficult to roll over

43:37 their existing stock of debt,

43:39 and there will be an increase in the default rates

43:41 and there will be an increase in corporate risk.

43:44 On top of that channel,

43:45 which is completely endogenous in the model,

43:47 there is also an exogenous type of mechanism

43:50 through which changes in sovereign risk

43:53 directly affect the corporate sector.

43:56 Although that is going to be in a reduced form way in my model,

43:59 this exogenous channel

44:01 can capture all the other channels that you may think

44:04 other than the bank lending channel,

44:06 for example,

44:06 a fiscal channel,

44:07 a trade channel,

44:08 and so on,

44:09 and inserted channel if you wish,

44:10 and so on.

44:11 And I'm going to use this extra layer of flexibility in the model

44:15 in order to target the estimates that I presented in my empirical analysis,

44:20 and I'm going to come back to that point.

44:22 Now,

44:22 the key ingredient that I introduced to this literature

44:26 is this two-way feedback loop

44:29 between

44:30 the network of the bank

44:31 and the corporate sector.

44:33 So the goal of this paper is to study

44:36 once we have this increasing sovereign risk,

44:39 how the presence of corporate risk

44:41 interacts with the financial

44:44 system as a whole with the balance sheet of the banks,

44:46 and how that interaction can significantly amplify

44:50 the effects

44:51 that were originated by the increase in the riskiness of the government.

44:56 So that is the model in a nutshell,

44:58 let me start describing the different sectors of this economy.

45:02 So then let's start with the heterogeneous firms.

45:05 Firms are going to be different across three dimensions

45:08 capital,

45:09 the long term loans,

45:10 and also in terms of their productivity.

45:12 OK?

45:13 What they're going to do is to use

45:14 their own stock of capital K.

45:16 They're going to hire labor in the market L

45:19 in order to produce the unique final good of this economy

45:23 using this

45:24 codalas type of technology.

45:26 Notice that this technology depends on two different productivity processes.

45:31 It first depends on this idiosyncratic productivity process

45:34 which follows an AR1,

45:36 OK?

45:36 And that process is exactly the same across all the,

45:39 the firms that I have in the model.

45:42 But on top of that idiosyncratic component,

45:44 it also depends on this same variable,

45:47 which is an aggregate variable,

45:49 and the idea is that this will allow me to capture

45:52 aggregate productivity losses

45:54 in the event

45:55 that the government defaults.

45:57 So the idea is that this sign will take a value of 1.

46:01 It is normalized to 1

46:03 if

46:03 the government is not in a default today,

46:06 but it will take a value of something smaller than 1

46:09 if the government is in a default today.

46:12 And

46:12 I'm going to use this type of approach,

46:14 of course,

46:15 a reduced form way

46:16 in order to exogenously link corporate risk to sovereign risk,

46:20 OK?

46:21 And I'm going to calibrate this particular variable

46:24 in order to match

46:25 the IV estimates that I presented in the empirical analysis.

46:33 Just coming in with a super quick,

46:34 super quick question,

46:36 uh,

46:37 why

46:38 it so

46:39 in,

46:39 in the case of Italy we never observe it

46:42 take the value of one,

46:45 but in Greece we do,

46:47 uh,

46:48 why not.

46:50 I,

46:51 I,

46:51 I know you're Italian,

46:52 but,

46:52 but,

46:52 but why not

46:54 include elements or,

46:56 you know,

46:57 even in a

46:59 reduced way,

46:59 uh,

47:00 uh,

47:01 the Greek.

47:02 Uh,

47:02 you know,

47:03 test this to bring data where you do get,

47:06 uh,

47:07 you know,

47:08 more action there.

47:09 Yeah.

47:10 Thank you,

47:10 Carmen.

47:10 That is an excellent comment.

47:12 I'm,

47:12 I'm from Argentina.

47:14 I'm not from Italy,

47:15 but I know that my last name may bring some confusion in that regard.

47:18 But let me tell you this,

47:19 uh,

47:20 I'm going to use this value to calibrate,

47:22 to match

47:23 the IV estimates,

47:25 but

47:26 this value that I'm using as part of the calibration of the of my model

47:30 is quite related.

47:32 To a lot of papers in the literature that

47:35 documents the drop in TFP upon a sovereign default.

47:39 So for example,

47:39 there is a paper by Guidos and Leis and

47:41 coauthors for Argentina in which they estimate this value

47:44 to be around 15%.

47:46 That is the drop

47:47 in TFP upon a sovereign default,

47:49 and that is a very similar value

47:52 to the one that I'm going to use

47:53 in order to match the IV estimates of the empirical analysis.

48:05 OK,

48:05 thank you.

48:06 OK,

48:07 please let me know if there is uh another question here.

48:09 Um,

48:10 all right.

48:11 So let me describe the recursive problem of these firms.

48:15 So

48:16 this is going to be the evaluation of the firm,

48:18 and it will depend on the

48:19 idiosyncratic states,

48:21 capital,

48:21 loans,

48:22 productivity,

48:23 and on the aggregate states which are captured with

48:26 this capital S vector.

48:28 So the owner of this firm is going to choose

48:32 the optimal next period stock of loans,

48:34 the optimal next period stock of capital,

48:37 in order to maximize

48:39 the current dividends

48:41 plus the expected continuation value of this particular firm.

48:45 And as I said,

48:46 this is a model with limited commitment,

48:48 so the firm

48:49 may default.

48:50 It has the ability to default on their loans.

48:53 And that default decision will depend essentially

48:56 on an outside option of the firm which is

48:58 captured by this red term or this VD function,

49:01 OK?

49:02 So the idea is that

49:03 to the extent that the continuation value

49:05 is smaller

49:07 than this exit outside option,

49:09 the firm will default on its loans and it will exit

49:13 the industry forever and it will be replaced by a new entrant.

49:17 Now,

49:18 this is of course going to be subject to a couple of constraints.

49:20 The first one is nothing more but

49:22 the definition of dividends.

49:24 So in this case,

49:25 the dividend of this firm is going to be the sum

49:28 of the current profits pi

49:30 minus the investment cost and there are a couple

49:32 of frictions behind this function that let me skip.

49:36 Plus the proceeds from issuing new loans.

49:39 So this queue is an endogenous subject in the model

49:42 and it is the unit price of each loan.

49:45 And this depends on two different things,

49:48 OK?

49:48 It depends on the riskiness of the firm,

49:51 meaning the default probability,

49:53 and it also depends on the balance sheet of the bank.

49:56 The idea is that

49:57 to the extent that the banks are constrained today,

50:00 this price will be smaller,

50:02 meaning that the spreads will be higher,

50:04 and that will have an implication

50:07 on the dividends of the firm,

50:08 on the evaluation of the firm,

50:10 and therefore on the incentive to default.

50:13 The terms in bracket is just whatever this firm is issuing,

50:16 OK?

50:17 In terms of new loans.

50:19 So this MF parameter is capturing the share of loans that are maturing today.

50:24 And the last term that you have here are the debt services,

50:27 OK,

50:27 so each period the firm has to repay

50:30 some coupons,

50:32 and also the share of loans that are maturing.

50:35 The last two are just,

50:36 uh,

50:37 this is imposing that dividends cannot be negative,

50:39 that is purely for simplicity.

50:41 And the last one is just the perceived law of motion

50:44 for all the aggregates of this economy.

50:47 The household side of the problem is going to be very simple and straightforward,

50:51 so I'm going to skip the details.

50:52 The idea is that

50:53 these agents are risk neutral.

50:55 They can consume the unique final good of the economy.

50:58 They can also save.

51:00 They only have one instrument available,

51:01 and those are short term deposits.

51:04 They will receive some lump sum transfer from the government,

51:07 and they are also the owners

51:09 of the bank of this economy.

51:11 Let me skip that recursive problem

51:13 and let me focus now on the government side of the economy.

51:17 So this is a model in which I'm not going to consider

51:21 the optimal default decision of the goal.

51:23 In that sense,

51:24 that would be exogenous.

51:25 So this is what I'm going to do.

51:27 Each period,

51:28 the government will issue some long-term bonds

51:31 in order to satisfy its budget constraint.

51:34 And that budget constraint depends on the profits of all the firms,

51:38 but it also depends on some exogenous component

51:41 basically that follows a fiscal rule,

51:44 OK?

51:45 And the idea is that

51:47 the price of these bonds,

51:48 it is also an endogenous object,

51:50 QBS.

51:52 And that object depends on two things

51:54 on the sovereign risk of this government,

51:56 meaning on the probability of default,

51:58 and it also depends on the balance sheet of the bank or

52:01 in the bank's stochastic discount fund.

52:05 Now,

52:06 how am I going to model

52:08 the default decision of the government which is exogenous?

52:12 This is following the Booda 2016 paper,

52:14 and the idea is that

52:16 in each period

52:17 there is going to be a realization of this HG variable which can take two values,

52:22 0 or 1.

52:23 If it is zero,

52:24 it means that the government is not going to default today.

52:27 If it is 1,

52:28 it means that the government will default today.

52:30 OK?

52:31 And this realization

52:33 depends on two additional things.

52:35 It depends on

52:36 some epsilon shock,

52:37 which is an IID type of shock.

52:40 But it also depends on the level of this S variable,

52:43 and I'm going to interpret that as

52:45 as sovereign risk.

52:47 And the idea is that

52:49 the larger the value of S,

52:50 then the more likely that the government will default

52:53 today for any given realization of this epsilon shock.

52:57 And this sovereign risk will follow this very simple A1 process,

53:01 and that is the only source

53:03 of aggregate uncertainty that I have in the model.

53:08 OK,

53:09 to close this economy,

53:10 I need to define the problem regarding the banks.

53:14 OK,

53:14 so remember,

53:15 the banks are owned by the households,

53:17 and the idea is that they're going to maximize

53:20 their respective net worth a

53:22 upon exit.

53:24 So W is the value of the bank

53:27 and what the banker is going to choose is

53:30 the optimal amount of government bonds that they want to hold,

53:33 the optimal amount of loans that they're going to

53:35 give to each of the firms in the model.

53:38 And the optimal amount of deposits that they're going to take

53:41 from the households,

53:42 OK?

53:43 And what they're going to do is to maximize

53:46 their expected net worth upon exit,

53:48 which is this expression that you have right here.

53:51 There are a couple of constraints,

53:52 of course.

53:53 The first one

53:54 is just the definition of the balance sheet.

53:57 So on the left hand side what you have are

54:00 the sum of the deposits of the bank

54:02 plus the net worth of the bank.

54:05 On the right hand side you have the market value of all the assets of this bank.

54:09 So the integral is

54:11 the market value of the loans to all the firms.

54:14 This last term is the market value of the government bonds that are being held

54:19 by this particular bank.

54:21 Now this constraint,

54:22 which I call the incentive constraint or also leverage constraint,

54:26 is the key equation that I have in the model.

54:28 OK.

54:29 This equation is saying that a fraction kappa

54:32 of the assets of the bank

54:34 cannot be larger

54:36 than the value of the bank w.

54:39 And why is this equation so important in my model?

54:42 Because this will allow me to link

54:44 endogenously sovereign risk

54:47 to corporate risk,

54:48 and this is the intuition behind it.

54:51 Suppose that there is an increase in sovereign risk.

54:54 Because

54:55 the banks are heavily exposed to that type of risk,

54:58 there will be a reduction in the net worth of the bank

55:01 and therefore in the evaluation of the bank,

55:03 so this W will decrease.

55:05 Now,

55:06 because of this incentive constraint,

55:08 to the extent that this inequality is binding,

55:11 the bank will have to react

55:13 by decreasing the loans

55:15 to the corporate sector.

55:17 So there will be a reduction in the credit supply or there will be an increase

55:21 in the spreads that they charge.

55:23 And that of course will affect

55:25 the incentive to default of the firms and therefore

55:28 the level of corporate risk.

55:31 The last two equations are just law of motions for next period net worth and

55:35 all the aggregates of the economy.

55:39 Let me skip this.

55:40 Let me jump down to the results that I have in this model.

55:43 So here we have some results for the non-stochastic steady state.

55:47 So this is an economy without

55:49 sovereign risk.

55:51 And what you have here is the firm distribution both by size,

55:55 that is the left panel,

55:57 and also by default risk,

55:59 OK?

56:00 So the bars are for the model.

56:02 The lines are for the data

56:04 and as you can see the model.

56:06 That's a good job in matching the two distributions.

56:10 This is somewhat somehow calibrated,

56:13 OK?

56:13 But this is completely untargeted,

56:15 OK?

56:15 So it is remarkable

56:17 that the model can capture this very asymmetric distribution

56:20 in terms of the default risk of the non-financial firms,

56:24 OK?

56:25 And I will come back to that point later on.

56:28 OK.

56:29 As a source of validation of the model,

56:32 what I'm going to ask is how well can this model

56:35 replicate the Italian crisis,

56:37 OK?

56:38 To address that question,

56:40 what I'm going to do is to fit the model

56:43 with a sequence of sovereign shock,

56:45 OK,

56:45 that resembles the one that you observed

56:48 in the Italian context,

56:49 OK.

56:51 The other three panels are endogenous variables in the model,

56:55 and as you can see,

56:56 the model does a good job

56:58 in matching the size

57:00 and the persistence of the crisis.

57:02 OK?

57:02 So the red line is the Italian data,

57:05 the black line is

57:06 the model in client dynamics.

57:08 Now,

57:08 it also does a relatively good job in matching the

57:11 increase in the leverage of the bank during the recession

57:15 and the posterior

57:16 decrease in banks leverage once the economy starts to recover.

57:21 And lastly,

57:22 even though this is sort of a targeted in the calibration,

57:26 the model does a good job in matching

57:28 the dynamics of corporate risk during the recession,

57:32 at least as implied

57:33 by my empirical estimates.

57:37 Yes,

57:37 there's a question from Tuan.

57:39 Yes,

57:40 yeah,

57:40 so,

57:41 so

57:42 there's something I didn't get.

57:42 So from your incentive

57:44 constraint,

57:45 the kappa thing.

57:47 Yes,

57:48 wouldn't the model generate multiple equibrium

57:50 because if enough firms default.

57:54 Then

57:55 a lot of firm will not be refinanced and will default.

57:58 We strategically default,

58:00 yeah,

58:00 so won't you get intrinsically multiple sensotic equilibria in this case,

58:04 yeah,

58:04 I mean

58:05 that type of concern,

58:07 uh,

58:07 is going to be present

58:09 whenever the pricing kernel,

58:12 sorry,

58:12 this QB.

58:14 is determined after

58:17 the firm's decision regarding how much debt they want to take

58:20 or after whenever they want to default or not.

58:24 In order to break that type of multiply equilibra concern,

58:28 this is the typical assumption

58:30 that this model of default they have

58:32 is that

58:33 first

58:34 the firm is going to choose whether to default or not

58:38 after you observe the default decision.

58:40 The price kernel of this function will be determined based on whatever you want to,

58:45 to borrow from the bank.

58:46 And that type of assumption allows you to break

58:49 this type of multiplicity of equilibrium.

58:57 Uh,

58:57 is that

58:58 OK,

58:58 OK,

58:58 perfect.

59:01 OK,

59:01 right,

59:02 sorry,

59:03 sorry,

59:03 there's one more question from Paolo.

59:06 Yes.

59:07 Yes,

59:08 so,

59:09 in the,

59:09 in the context of the Eurozone crisis,

59:12 um,

59:13 Italy was,

59:13 was thought to be too large to fail.

59:16 So what happened to the real economy in Italy and to the debt

59:19 crisis in Italy would also determine the institutional decisions at the EU level.

59:25 How,

59:25 how,

59:25 how,

59:25 how can you think about that in the,

59:27 in the,

59:27 in the model here?

59:29 OK,

59:30 so basically that will be related

59:32 to one of the

59:34 policy analysis that I'm going to consider.

59:36 So because there was this concern regarding

59:39 these banks are too big to fail,

59:41 I'm,

59:42 I'm going to consider some policies in

59:43 which the Italian government was injecting capital

59:48 towards these banks,

59:48 and I'm going to

59:50 analyze in the context of my model

59:52 what are the benefits of that particular policy.

59:54 Of course,

59:55 this model won't be able to say anything at

59:57 all regarding moral hazard because those types of policies,

1:00:00 they do have

1:00:01 an important type of moral hazard component,

1:00:03 but I will be able to quantify the effects of that particular policy

1:00:07 in the context of the Italian economy.

1:00:13 Sure.

1:00:15 Yeah,

1:00:15 I mean,

1:00:16 in the

1:00:18 bank um

1:00:20 lending channel that you're considering,

1:00:21 you're looking at the asset side of the bank,

1:00:25 um,

1:00:26 but the liability side of the bank can also be hit.

1:00:28 Um,

1:00:29 uh,

1:00:30 there is some evidence

1:00:32 that the banks usually have a higher withdrawal of deposits.

1:00:36 When you have more links to sovereign risk,

1:00:39 so that contraction in the deposit base also

1:00:42 affects the lending capacity of the bank.

1:00:45 That seems like you're kind of ruling out that channel.

1:00:47 Yes,

1:00:48 uh,

1:00:48 for sure,

1:00:48 I,

1:00:48 I do consider that channel to be at play,

1:00:51 at least in this very large Italian crisis.

1:00:54 Uh,

1:00:54 as you said,

1:00:55 uh,

1:00:55 this model is not going to be able to say anything regarding that.

1:00:59 Of course,

1:00:59 the type of policy implication that you may have in a model subject to bank

1:01:03 ranks may be a little bit different to the ones that I'm going to present next

1:01:07 because in that case you have more incentive to

1:01:09 recapitalize the banks in order to avoid a potential bank

1:01:12 run,

1:01:13 but that is something that I'm not considering so far,

1:01:15 but

1:01:15 obviously it is very important to add that for future research in one of my papers.

1:01:22 OK.

1:01:23 Go ahead.

1:01:25 Uh,

1:01:26 I think one very brief observation is,

1:01:29 you know,

1:01:29 the,

1:01:29 the,

1:01:30 the narrative your,

1:01:31 your,

1:01:31 your model

1:01:32 gives us is

1:01:35 grossly oversimplified,

1:01:36 but it's a common creditor channel in which you,

1:01:39 you have contagion from the sovereign to the,

1:01:42 to the,

1:01:43 uh,

1:01:43 corporates through the common creditor in this case,

1:01:46 the banks,

1:01:47 you know,

1:01:47 there is an analogous literature on cross-border contagion also,

1:01:52 uh,

1:01:53 connected to the common creditor,

1:01:55 but you know,

1:01:56 it,

1:01:56 it,

1:01:56 it,

1:01:57 it really,

1:01:58 I,

1:01:58 I mean,

1:01:58 and it's simplest stripped down version.

1:02:01 I think the common predator story is something perhaps you could,

1:02:05 you know,

1:02:06 um.

1:02:07 Emphasize more.

1:02:10 I see,

1:02:11 I see.

1:02:12 I do agree with you,

1:02:13 and my contribution to that broader type of literature

1:02:16 is to show

1:02:18 the importance of the heterogeneity

1:02:20 in the non-financial firms,

1:02:22 meaning in the borrowers,

1:02:23 and that is something that I think that it will be clear

1:02:27 with this picture.

1:02:29 So what you have here

1:02:30 are the IV estimates

1:02:32 for of the empirical analysis

1:02:34 that is in black,

1:02:35 OK?

1:02:36 And you have the model implied estimates

1:02:39 and

1:02:39 based on simulations of the model.

1:02:42 Now,

1:02:42 as I said,

1:02:43 I'm using this

1:02:45 reduced form type of productivity losses in

1:02:48 order to target this particular estimate,

1:02:50 OK?

1:02:50 This IV estimate.

1:02:51 This is a targeted moment in my calibration.

1:02:54 All the other IV estimates are completely untargeted in the mall.

1:02:59 So the model does

1:03:00 a really good job in matching this

1:03:03 heterogeneous reaction to increases in sovereign risk.

1:03:06 The idea,

1:03:07 as I said at the beginning,

1:03:08 is that

1:03:09 the safer firms are almost unaffected by the increase in sovereign risk,

1:03:14 but the riskier firms are significantly more affected.

1:03:18 So in terms of the effects

1:03:19 of the corporate sector

1:03:21 in the financial system,

1:03:23 The point that I want to make with this

1:03:25 is that there is going to be an

1:03:27 amplification mechanism going from the increasing corporate risk

1:03:31 to the non-final to the financial system.

1:03:33 But that amplification mechanism

1:03:35 is driven purely

1:03:38 by these very risky firms.

1:03:41 OK.

1:03:41 The majority of the firms,

1:03:43 they are not going to observe a very large increase in their corporate risk

1:03:47 after a sovereign shock.

1:03:49 So those firms are not going to amplify

1:03:52 all the effects.

1:03:53 The problem is coming from this subset of riskier firms

1:03:56 which are going to default more,

1:03:58 and that will have an important impact on the balance sheet

1:04:01 of the bank and in the economy as a whole.

1:04:03 OK.

1:04:06 So I do have 10 more minutes.

1:04:08 So let me just

1:04:09 summarize in 2 more minutes the

1:04:12 decomposition of all the effects that are at play in the model.

1:04:15 OK?

1:04:16 So what you have here

1:04:17 is the impulse response to a 3 standard deviation increase in sovereign spreads,

1:04:23 OK?

1:04:23 And you have 6 different endogenous variables in the model GDP,

1:04:27 bank's net worth,

1:04:28 capital spreads,

1:04:29 and so on.

1:04:30 The line in black is capturing the full effects in my model,

1:04:34 OK?

1:04:34 When all the channels are active.

1:04:37 Now,

1:04:37 in the first counterfactual that I'm going to do.

1:04:40 is to analyze

1:04:42 the effects

1:04:43 that are coming from the contraction in the supply of credit

1:04:47 only due

1:04:48 to the increasing sovereign risk,

1:04:50 OK?

1:04:51 What I'm going to show is that under this counterfactual,

1:04:55 You can explain only a small fraction of the crisis,

1:04:58 but that is consistent with the previous literature.

1:05:03 So,

1:05:03 sorry.

1:05:04 So here you have this counterfactual.

1:05:05 This is the one in red.

1:05:07 So as I said,

1:05:08 this is a world

1:05:09 in which corporate risk is almost constant

1:05:11 and all the effects are coming from the

1:05:14 increasing sovereign risk and how that increase affected

1:05:18 the supply of credit.

1:05:21 You can see that under that scenario you can only explain

1:05:24 about 20% or even less of the total drop of GDP,

1:05:28 so only a small fraction of the crisis.

1:05:30 Now what happens when you move to a world in which you consider corporate risk?

1:05:36 I will show you that you get a much larger drop in output.

1:05:39 Of course behind this larger drop there are two things going on.

1:05:43 You have the exogenous mechanism,

1:05:45 OK,

1:05:46 but you also have the indulgence mechanism which is the bank lending channel.

1:05:49 Let me show you that

1:05:51 under this war with corporate risk,

1:05:53 my endogence mechanism,

1:05:55 the bank lending channel,

1:05:56 can account for 40% of the decline,

1:05:58 and from this 40%,

1:06:01 about half of this

1:06:02 is explained

1:06:03 by corporate risk.

1:06:05 OK?

1:06:06 So this is the control factor that I'm going to consider,

1:06:08 the one in blue.

1:06:10 So this is a world in which changes in sovereign or corporate risk

1:06:15 are not going to affect

1:06:17 the network of the bank.

1:06:18 So

1:06:19 the network is

1:06:20 constant in the simulations.

1:06:22 So all the effects that you observe

1:06:25 in this blue line

1:06:26 in this second panel

1:06:28 are completely exogenous in the model and are

1:06:30 coming from this reduction in firm's productivity,

1:06:33 OK,

1:06:34 so the difference between my baseline model,

1:06:36 which is the black line,

1:06:37 and this other control factor in blue,

1:06:40 so this shaded blue area.

1:06:42 This is capturing the share uh explained by

1:06:46 my indulgence channel or

1:06:48 by

1:06:49 my bank lending channel.

1:06:50 So this is about 40% of the crisis.

1:06:53 Behind that decline,

1:06:55 there are two things going on.

1:06:56 You have

1:06:57 the effect of sovereign risk on the balance sheet of the bank,

1:07:00 but at the same time you have

1:07:02 the effect of corporate risk on the balance sheet of the bank.

1:07:06 So to understand the importance of corporate risk,

1:07:09 this last counterfactual highlights the additional drop that you get in GDP

1:07:14 once you factor in

1:07:17 how this increasing corporate risk

1:07:19 affects

1:07:20 the balance sheet of the bank.

1:07:22 So this gray area that you have here

1:07:24 is saying that at the peak of the crisis,

1:07:27 about half of this mechanism

1:07:29 is explained by corporate risk.

1:07:32 And as the economy recovers,

1:07:35 the fraction is almost 100%,

1:07:37 OK?

1:07:38 And this type of effect was completely

1:07:41 missing from this sovereign debt literature.

1:07:43 And the main takeaway from the model is that

1:07:45 You have a sovereignty crisis,

1:07:47 the financial crisis,

1:07:49 but you still need to carefully consider

1:07:51 what is happening with the non-financial firms

1:07:54 because these firms are going to significantly amplify

1:07:58 whatever is happening with GDP

1:08:00 through

1:08:01 their effects on corporate risk on

1:08:04 the balance sheet of the bank.

1:08:05 And that's why I said at the very beginning

1:08:08 that corporate risk can significantly amplify both the size

1:08:12 but also the persistence of a sovereign debt crisis.

1:08:15 And that is the new result that I'm bringing to this literature.

1:08:20 OK.

1:08:21 So,

1:08:22 let me skip this.

1:08:23 Um,

1:08:24 let me,

1:08:24 in the last 6 or 7 minutes that I have

1:08:28 described the different fiscal policies that I'm going to consider,

1:08:31 OK?

1:08:33 So because I won't be able to analyze

1:08:36 how changes in the fiscal deficit

1:08:39 have an impact on the sovereign risk,

1:08:41 this is the type of exercise that I will do.

1:08:44 I'm going to consider 4 different fiscal policies,

1:08:48 with the same fiscal cost,

1:08:50 and I'm going to see

1:08:51 which of these policies

1:08:53 delivers

1:08:54 a larger reduction in the drop of GDP during the crisis.

1:08:59 So the first two policies are policies in

1:09:01 which you are helping directly the corporate sector.

1:09:04 Policy number one is one in which you are

1:09:07 basically giving a debt relief

1:09:10 to all the firms in the economy.

1:09:12 Policy number 2 is one in which you are going to give this

1:09:15 debt relief but only to the riskier firms,

1:09:18 meaning firms with a default probability

1:09:21 higher

1:09:22 than a certain threshold.

1:09:24 And the last two policies are more in line to what we observed

1:09:28 during the last European debt crisis.

1:09:31 For example,

1:09:32 policy number 3

1:09:33 is

1:09:34 the one in which the government is repaying all the non-performing loans.

1:09:39 Policy number 4 is the one in which

1:09:42 the government is injecting capital

1:09:45 directly

1:09:46 into the financial system.

1:09:48 OK?

1:09:49 What I'm going to show is that

1:09:51 the debt relief program

1:09:53 towards the riskier firms.

1:09:55 Displays very important efficiency gains,

1:09:58 OK?

1:09:58 And let me give you the intuition and then I will show you the results.

1:10:02 The intuition is the following.

1:10:04 When you're helping these riskier firms,

1:10:06 you can achieve two different goals at the very same time.

1:10:10 On the one hand,

1:10:12 Because

1:10:12 of course you are helping those firms closer to the default boundary,

1:10:16 you are going to obtain a very large drop

1:10:19 in the number of corporate defaults.

1:10:21 OK?

1:10:22 That is pretty obvious.

1:10:24 But now at the same time by doing this,

1:10:26 you are decreasing the overall level of corporate risk in the economy

1:10:31 and because banks are exposed to that type of risk,

1:10:34 this type of policy allows you to indirectly recapitalize the banks

1:10:38 and that of course is going to benefit all the firms in the economy.

1:10:43 OK?

1:10:44 So let me give you the results in the last

1:10:47 5 or 4 minutes that I have.

1:10:49 So

1:10:50 this is the fiscal cost for all these 4 policies

1:10:54 and the policies that I want you to focus on is the one that is in blue.

1:10:57 That is the debt relief

1:10:59 towards these riskiest firms,

1:11:01 OK?

1:11:03 So in terms of capital or also in terms of the change in GDP,

1:11:06 notice that this policy,

1:11:08 the one in which you are helping the riskier firms,

1:11:11 allows for

1:11:12 a smaller contraction in GDP or in capital during the recession,

1:11:16 OK?

1:11:16 And the

1:11:17 the difference seems to be significant

1:11:20 according

1:11:20 with respect to the other three policies.

1:11:23 So why is this policy better?

1:11:25 The intuition is the one that I was giving you before.

1:11:28 Notice that

1:11:29 the change in the default rate under this policy

1:11:32 is significantly smaller than

1:11:35 the increase that you observed

1:11:37 in

1:11:37 the other three policies or under the

1:11:39 baseline and scenario grade without any policy,

1:11:42 OK?

1:11:43 And the idea is very intuitive.

1:11:44 You are helping the firms

1:11:45 that need it the most,

1:11:46 OK?

1:11:46 That's very obvious.

1:11:48 Now the key point that I want to emphasize with this paper,

1:11:51 this is the key ingredient that I want to convey

1:11:53 is that

1:11:54 this policy which you are helping the non-financial firms.

1:11:58 Can have a very positive impact on the network of the bank,

1:12:02 OK?

1:12:03 So that is something that you can observe in this last panel.

1:12:07 So the policy in which we are helping the riskier firms

1:12:10 is the 2nd to last bar that you observe here.

1:12:13 And notice that the reduction

1:12:15 in the network of the bank

1:12:17 is very similar

1:12:18 to a policy in which you are directly injecting capital towards the bank.

1:12:23 So the amount of the reduction in the network is

1:12:26 very similar in the sense that you're going to get

1:12:28 all the benefits from this policy,

1:12:30 but at the same time,

1:12:32 you are saving this policy,

1:12:33 these firms from a default,

1:12:35 which is a good thing.

1:12:37 Now,

1:12:38 given that you are able to recapitalize the banks,

1:12:40 you are going to observe some positive spillovers

1:12:44 towards firms with small

1:12:46 or medium levels of risk which are not

1:12:49 directly targeted by this type of intervention.

1:12:52 And that is something

1:12:53 that heterogeneity is something that you can absorb

1:12:56 in this last figure that I have for today.

1:12:59 So the first panel shows the change in the

1:13:01 default rate for firms with different levels of risk,

1:13:05 safe,

1:13:05 medium,

1:13:05 and risky firms.

1:13:07 The bottom panel shows changes in

1:13:10 the capital stock of these firms for safe,

1:13:12 medium,

1:13:12 and risky firms.

1:13:14 So this policy in blue is targeting only a subset of the risky firms,

1:13:19 OK?

1:13:19 Mostly half of these firms are being targeted by this intervention.

1:13:23 And of course,

1:13:24 by doing this,

1:13:25 there is going to be a large reduction in the default rate and also

1:13:28 a large contraction in the drop of capital that is kind of obvious.

1:13:32 But the key idea is that

1:13:34 once you do this,

1:13:35 you are also recapitalizing the banks

1:13:37 and that will

1:13:38 create a benefit

1:13:40 on firms that are not directly targeted and that is something that you can Observed

1:13:44 in these

1:13:45 two panels right here,

1:13:47 so the contraction in capital

1:13:49 is going to be

1:13:50 dampened for the medium and for the safe firms because

1:13:54 given that the banks are more capitalized,

1:13:57 they will be able to have access to cheaper credit or to a larger

1:14:01 credit supply.

1:14:03 So I know that I'm running out of time,

1:14:05 so just let me conclude.

1:14:08 In the empirical analysis,

1:14:09 I use Italy purely as a case of a study during the last European crisis

1:14:14 using an identification strategy that I

1:14:16 exploited

1:14:17 the volatility of Italian sovereign risk around some events.

1:14:22 I showed that sovereign risk can account for almost 30%

1:14:26 of the total increase in corporate default risk

1:14:29 and also,

1:14:29 and this is my contribution to that literature,

1:14:32 that there are very important

1:14:33 asymmetries across France

1:14:36 behind that transmission.

1:14:37 Then I provide a very simple type of exercise in

1:14:40 which I show that the band lending channel seems to be

1:14:43 an important driver behind the transmission and that

1:14:45 is consistent with a very large empirical literature.

1:14:49 Then I took these two facts to motivate a quantitative model

1:14:53 in which I showed that corporate risk

1:14:55 significantly amplified both the size and the persistence of the crisis,

1:14:59 and the idea is that this increase in corporate risk

1:15:02 will further affect the bank's balance sheets and that will contract even more

1:15:07 the credit supply.

1:15:09 Lastly,

1:15:10 consistent with my Pican analysis,

1:15:12 I showed that the riskier firms are more affected by increases in sovereign risk.

1:15:17 So I studied different policies that exploit that heterogeneity

1:15:21 in order to provide some policy recommendation,

1:15:23 and what I show is that those policies that

1:15:25 are geared toward the riskier firms are more effective.

1:15:28 Because

1:15:29 they present positive spillovers

1:15:32 through

1:15:32 the interactions with the financial system and those positive spillovers

1:15:36 end up benefiting the safer firms of them all.

1:15:39 As the next step is

1:15:41 to introduce some macroprudential policies in this context

1:15:44 to assess the benefit of those policies.

1:15:48 Thank you very much,

1:15:49 everybody.

showAllTimestamps

no

transcript

OK. So thanks everybody for uh joining in. Like, uh, we have another job market seminar today from uh someone who is a familiar face to some of you, because Matthias Moretti used to work at the bank before he started his PhD and he's going to present to us a very interesting paper on the asymmetric posture of sovereign risk. So Matthias, take it away. You have 75 minutes and for the 1st 10 minutes, we're gonna respect the no questions rule. OK, OK. Thank you very much, Bob, for that introduction. So this is the title of my job market paper, The Assymetic Pathrough of Sovereign Risk, and thank you very much for, for this opportunity and for attending my presentation. So let me jump directly to the motivation of my model. And let me start saying that it is a very well known empirical fact that sovereign debt crises are characterized by very large and persistent declines in economic activity. So for example, think about Italy or Spain. It took more than 7 years for their economies to fully recover. Now, the question clearly here is why is that the case? Why are these crises so persistent? One of the most common explanations that you will find in the literature is based on the exposure of the domestic banks to sovereign risk, and the idea in all these studies is that the domestic banks typically hold a very large share of government bonds, and because of that, whenever there is an increase in sovereign risk that will weaken banks' balance sheets, banks will tighten the supply of credit, and that propagates the effects to the rest of the real economy. Now in my market paper what I show is that behind that explanation there is a very important missing piece, and that is corporate risk. So I'm going to show today that sovereign debt crises are characterized not only by increases in sovereign risk but also by very large increases in corporate risk. In the empirical analysis, I will address two different questions. The first one is to what extent is this increase in corporate risk driven by the initial increase in sovereign risk? And also I'm going to shed some light on the different channels that are driving that particular relationship. Now the key issue here is that the domestic banks are of course heavily exposed to this corporate risk, and because of that, this additional increase in corporate risk will further affect the balance sheet of the banks and banks are going to contract even more the credit supply. So to understand that concern or that issue, I'm going to build a quantitative model in order to study and quantify the role of corporate risk as an amplification mechanism of a sovereign debt crisis. In terms of empirical analysis, the first thing that I'm going to do in this paper is to measure the transmission from sovereign risk to corporate risk. To address that, I'm going to use Italy as a case of a study and using this heteroschasticity-based approach that exploits differences in the volatility of Italian sovereign risk around some news events that I will describe. What I find is that sovereign risk can account for almost 30% of the total increase in corporate risk. And on top of that overall effect, there are very important heterogeneous effects across the non-financial firms. In particular, what I find is that the riskier firms, meaning those firms with a higher default probability, are the ones that are more affected by this sovereign shock. In the second part of my empirical analysis, again using Italian bank level data in this case, I'm going to shed some light on the different channels that are behind this transmission of sovereign risk to the non-financial firms. And although there are many, many channels behind it, of course, what I do in this section is to study and quantify the importance or the role of the domestic banks in this particular transmission. So I'm going to present some evidence that shows that not surprisingly, the banks are passing on sovereign risk to their corporate clients. In particular, the result that I have here shows that those banks with higher sovereign exposure before the crisis, meaning those banks holding a larger amount of government bonds, those are the ones that tend to exhibit. A larger increase in their corporate and non-performing loans. Now, in the second part of the paper, and this is my main contribution to the literature, I'm going to ask to quantify the importance of corporate risk as an amplification mechanism of a soaring debt crisis. And the key idea is that there is going to be an interaction between the sovereign risk, the corporate risk, and the financial system as a whole. To address that type of method, there are 3 key ingredients or elements in my model. So first, motivated by these empirical results, I'm going to feature a model in which firms are heterogeneous and in which they can default on their loans. The banks, the domestic banks, are going to provide all the loans to those firms and at the same time they're holding government debt, and that's why these banks are exposed to both sovereign and corporate risk. I'm going to assume certain financial frictions that I will describe and also this is a model with aggregate uncertainty, and the only source of aggregate shock is what I call a sovereign risk shock that I will explain in a second. Now the key mechanism in the model is that under these three key ingredients, corporate risk is endogenously linked to sovereign risk through the banking sector, and I'm going to use the empirical estimates of the previous analysis in order to discipline this particular relationship. So the idea here is that if there is an increase in sovereign risk, the balance sheet of the bank will be affected and the bank's lending capacity will be affected. So banks will charge higher interest rates, and in that context, given that firms can optimally choose when to default, more firms will have more incentive to default, and that increases the default rate, but also corporate risk, and that extra increase in corporate risk will further affect the balance sheet of the bank. And because of that, The model delivers a two-way feedback loop between the corporate risk of these non-financial firms and the balance sheet of the domestic banks, and I'm going to show you during this presentation that this feedback loop significantly amplies both the sense and the persistence of a sovereign risk shock. Finally, consistent with the empirical findings, the model can capture the asymmetric transmission of sovereign risk to the non-financial firms. So I'm going to use the model. I'm going to exploit this heterogeneity in order to provide some policy guidance. OK, so where does my paper stand with respect to this sovereign debt type of literature? There are essentially 3 different categories here. So the first category tend to view the fundamentals of the economy as given and they study how changes in these fundamentals affect the government's incentive to default and therefore sovereign risk. Now a second strand takes the opposite approach. They tend to view sovereign risk as an exogenous shock, and they study the channels through which changes in sovereign risk affect the real economy. In terms of empirical analysis, the closest paper to mine is this one by Heber and Schreger for the Argentine case. In terms of the quantitative model, my paper is mostly related to this paper by Bocola 2016. Now there is a 3rd strand of course that I study the feedback effects between the real economy or the non-financial firms and the sovereign risk. For instance, this paper, new paper Barraano Bay and Bocola. So where can I classify my papers in these 3 different categories? Essentially in the second one. So what I'm going to do is to view sovereign risk as an exogenous shock. I'm going, I'm going to contribute to this literature because I'm going to explicitly model the behavior of corporate risk during a sovereign debt crisis, and I'm going to study and quantify how the presence of corporate risk through the interactions with the financial system can further amplify the effects of a sovereign risk shock. OK. Let me pause here and please let me know if you have any questions. If not, I will move to the empirical analysis. Claudia has a question. Claudia, I want you go ahead. Hi, Maths. Uh, so I have a question here. So, uh, so you're saying that banks with more sovereign debt, uh, is, are gonna tighten their margins to non-financial firms. Uh, so I'm thinking on, uh, my work on Mexico. So in Mexico, I see that usually banks that lend more to the government also are more likely to lend to firms, non-financial firms that are more connected. To the government, so suppliers of the government. So then, could this tightening that you observe or how would this enter in your model, that this tightening is not coming, you know, from banks tightening credit supply to a random firm, but to, to firms that were also exposed to, to this uh sovereign shock of the government because they are suppliers of buyers or buyers of the government. Yeah, thank you, Claudia. That is an excellent question. So you will see that through the lens of my model, all there is going to be one representative bank and the supply will be affected because the balance sheet of the bank will be affected. But in that sense, my model will not be able to capture, for example, firms that rely more or less on public subsidies in order to operate. So that will be completely outside the scope of my paper. Now in terms of the literature that is already there, at least for the Italian case, there is a Paper by Botero Lenzu and co-authors in which they show that the same firm that is operating with two different banks, the type of bank that is highly exposed to this sovereign risk, contracts the supply even more relative to the other bank to this particular firm. So even after you control for all these firm level characteristics, the type of bank lending channel that I have considered in the model is already there in the literature, and there are a lot of papers that document that type of relationship. Alvaro Please go ahead and ask your question. Uh, yeah, thank you, uh, about, uh, so Matthias, one quick, quick question. So I mean, naturally when there's, uh, sovereign risk or where when trade agencies downgrade the sovereign, uh, sovereign, uh, all or the private is somewhat, you know, benchmark against the, the government rating. So naturally when the rating of the government goes is lower than. All firms in the country will be downgraded or this the direct effect of, of, you know, more costly is that that what you have in mind here or not necessarily because you're, yeah, thank you, Alvaro. Uh, this is not exactly the type of mechanism that I have in mind. I know that there are a couple of papers that exactly exploit that type of of of of credit ratings between the corporate and the sovereign government because there is kind of like a ceiling in which the corporates cannot be graded above the sovereign. That it will not be present at least in my empirical or quantitative model, but I do agree that it's another channel through which the changes in sovereign risk may end up affecting the corporate risk, and that is a channel that works completely outside the bank lending channel. And I agree with you that may be present in the effects that I'm going to capture today, but that is not what I'm going to use in terms of the identification of the effects. I'm going to rely, as I'm going to explain it on some news that are coming from other countries. OK. OK, so I think you were as well. Oh sorry, Sergio, go ahead and ask your question. Um, thank you, Mathias. One question, you place your, your paper more on the second, uh, strand of the literature, but by increasing corporate risk and affecting the real economy, you can also hit, uh, the sovereign risk. Easily. Uh, I, I understand that you might not capture that in your model, but that's, that would be something that will affect the sovereign as well. Yes, uh, as, as you said, in my model, I'm not going to consider purely for simplicity, of course, the feedback effects between the real economy and the sovereign risk of the government. But you will see that I'm going to calibrate the model in order to target the causal effect going from sovereign risk and corporate risk. So I'm going to estimate that on the empirical analysis, and I'm going to use that in order to calibrate the model because of course my model won't be able to say anything at all regarding this feedback loop between the fundamentals of the economy and sovereign risk. Art Um, hi, Matthias. Uh, uh, uh, you can postpone this to the empirics if you, if you prefer, but, um, I'm just wondering how you're going to, uh, distinguish, um, spillovers from, from sovereign risk to corporate risk from, you know, very standard kind of common shocks that are going to be going on. Like, you know, uh, depreciation of the exchange rate is going to be correlated with sovereign risk and have a standard balance sheet channel effect on, uh, on, uh, Yes, yes, let, let me postpone that discussion because that is the main identification concern, and I will be, I will come back to that point because it's one of the most important points in the empirical identification strategy. Fun, go ahead. Yeah, one clarification, so you're, you're considering firms in the country for which government bonds have been downgraded? Is that the, is, is it going to be a, a, a country-level analysis or do we consider also all the countries that own sovereign bonds, on the bonds of a given country for which it's downgraded? Uh, this is going to be a country-level type of analysis because I'm going to focus on Italy only as a case of a study. Uh, and the idea is that I'm going to use some news coming from other countries in order to identify the effects, but at the end of the day, this is a result that is based on the Italian economy. Yeah, so my question, for example, if a lot of countries own US debt, yes, that doesn't apply, that would not apply to the US in a sense. It has to be the case that a lot of the debt is owned by domestic, exactly. And, and in Europe as a whole, the domestic banks in Italy are holding a significant share of the Italian government bonds. Of course in the US that is, that is different for sure, yes. But this is not specifically something regarding Italy because it's something that you usually tend to observe in other European countries, Spain, Greece, Ireland, and so forth. Uh, now, I, I don't want to delay you further, but Will you be discussing in the, in this context also the role of regulation or moral suasion because in Italian banks, as you know, you know, and especially during the height of the Greek crisis, the government was Basically lining up the banks telling them to participate in auctions. So you know, there's an element whether you want to call it, you know, in that particular example, moral suasion or or regulatory changes that get the banks to buy the more debt. That, I mean that is an excellent question. That will be completely outside the scope of my paper. I know there are papers, for example, by Ivain in which they explicitly study this concern regarding moral suasion, and that is something I'm not going to consider, at least in this version of the paper. OK, great. Darren, no more questions for now, so feel free to continue. OK, OK, thank you. So let me jump to the empirical section. I will try to be brief here, at least to highlight the the identification strategy and the main results, but I would like to emphasize more the second part of the paper, which is my quantitative model. So I will give you at least some highlights or not the details of what I did, but just the big picture. OK. So as I said, uh, I'm using Italy purely as a case of a study during the last European debt crisis, and I have two goals in this empirical section. The first one is going to estimate this effect going from sovereign to corporate risk, and that will be a central part of my analysis because I'm going to use those estimates later on to calibrate the quantitative model that I will be presenting next. As a 2nd goal, I'm going to highlight at least the importance of the domestic banks in this transmission of sovereign risk to the non-financial firms. Of course, in order to address those two goals, the first thing that I need to do is to construct a measure of corporate risk. And you will see that the type of identification strategy that I'm using requires high frequency data, basically daily data, and for that reason, I'm going to compute a proxy of corporate risk following the distance to default approach by Merton. The idea is that I have a panel of 120 publicly traded non-financial Italian firms, and this method allows you to combine the daily stock price together with some annual balance sheet level data in order to compute the risk neutral default probabilities of each of the firms that I have in my sample, and that's what I'm going to interpret as a good proxy of corporate risk. Now, why I'm using this measure and not other measures such as changes in the CDS spread or the corporate spreads of the bonds? Because by doing this, I can have access to a larger pool of non-financial firms. And because I want to describe this heterogeneity across the non-financial firms, having a larger pool of firms is a central element of the analysis. OK. Let me skip the details of this merchant approach. Let me just jump to the results that I have. So in this figure, you have this measure of corporate risk that is in red based on the merchant framework and in the black line, you have the measure of sovereign risk based on the 10 year CDS spread of the Italian government. Now, as you can see, it is very clear, there is a positive correlation between the two measures and as the crisis unfolds between 2010 and 2012, there is not only an increase in sovereign risk but a very large increase in corporate risk. And the key thing that I want to emphasize here is that behind these overall trends or these aggregate trends, there is a very important degree of heterogeneity across the non-financial firms, and that is something that you can observe in this picture. So here you have the same measure of corporate risk. But the black line shows the median level of this risk for the panel of firms that I have and the gray area shows a measure of the dispersion. So this is the inter-quantum range. So what you can see is that as the crisis develops, there is not only an increase in the median level of risk, but also a very large increase in the dispersion of this risk. So this highlights at least that there is a very important heterogeneity across the non-financial firms during this sovereign debt crisis. OK. Going now to the, if I may just quickly interrupt, uh, how do we know really it's the sovereign debt crisis, right, because there's also a business cycle undergoing this, and, uh, you know, the spreads between high yield corporates. And high grade corporates widens during downturns, uh, economic downturns, you know, has a big cyclic, uh, component, and here we're just looking, you know, at, at, at one part of the cycle. No, no, I, I completely agree with you. I mean this is just to highlight this asymmetry, uh, and now with identification strategy using high frequency data at the daily frequency, I will be able to identify that heterogeneity that is coming from sovereign risk. This is just to give you the big picture and nothing more than that. OK. OK, so going now to, to the identification strategy, what I'm going to do is to estimate this effect going from sovereign to corporate risk. Now there are obvious challenges when you want to do this. The first one is related to causality because it may be the case that the crisis started in the non-financial sector and then propagated to the sovereign risk of the Italian government. Another concern, as Art was mentioning at the very beginning, is regarding common factors. So this is a period in which there was a lot of volatility in Europe as a whole. So there may have been an increase, for example, in risk aversion or an increase in the market price of risk that is driving at the same time both sovereign and corpo risk. In order to address those types of concerns, this is the identification strategy that I proposed in this paper. So I'm going to rely on a set of what I call 4 news events that will allow me to capture in a sense, the exogenous variation in Italian sovereign risk. So these 4 news events are news that are coming from two other European countries, Greece and Portugal, and I have two different sets of news. I have news about bailouts for these 2 other economies and also news about some credit rating downgrades. The idea is that this news will have an effect on the Italian sovereign risk that is not related to the fundamentals of the Italian economy, and I will come back to that point. So overall, I have a set of almost 30 events between the three big years of the crisis, 2010, 202011, and 2012. So the question is, OK, I have this set of events. How am I going to use those events to estimate the effect going from sovereign to corporate risk? Formally, what I'm going to do is to use this heteroschasticity-based approach by Rigo Bonanzak. And the idea of this approach is that I'm going to exploit the differences in the volatility of Italian sovereign risk around these foreign news events. So this same identification strategy is the one used by, for example, Heber and Schreger in the context of the Argentine sovereign debt crisis, but it's also widely used in the monetary policy type of literature. Rigobonazaki is one example. Another example is this paper by Nakamura and Steinson. So let me give you at least uh a brief summary of what is all this method all about. OK. And in order to do that, let me first describe the set of events that I have captured with this foreign news coming from Greece and Pug. On the graph on the left, on this scattered plot, what you observe is the change in Italian sovereign risk. On the Y axis, you have the change in Italian corporate risk. The red dots are for all these 29 events that I have. The blue dots are for all the other days that I have in the sun. It is clear from the picture that there is a positive correlation between the two of these measures, not only during the event days but also during the non-event dates. Now, the table that you have on the right displays some moments around a 3 day symmetric window around these 4 news events. The only thing that I want to highlight from this table is this row that is highlighted in blue. So notice here that the changes in the volatility, sorry, in changes in Italian sovereign risk is higher during the event days compared to the non-event days. And you can show that that difference is strongly significant. The key idea of this Rigobon and Zack type of approach is to exploit the differences in these volatility. So let me Now, go to the details of this particular framework. Consider the following very simple system of equations, OK? In which the change in corporate risk is just a linear function of whatever change is happening in Italian sovereign risk, some common factors X and some shocks, OK? And the same goes for the change in Italian sovereign risk. In this case, this is the function of corporate risk, some potential of certain common factors, and some shocks. Now, what I'm going to do is to classify all the days that I have in the sample between the event and the non-event days using these four news events. And the formal identifying assumption behind this approach is that the variance of this a shock, meaning the shocks to Italian sovereign risk, is significantly higher during these foreign news events, and there is a whole literature of sovereign contagion that shows and provides evidence that backs up this type of assumption, OK. But now, the second part of the, of the assumption is that the variant of this epsilon shock, meaning the shocks to Italian corporate risk and also the variance of this potential observed common factors remain the same, OK? Under those assumptions, the exclusion restriction can be interpreted as follows. The idea is that the foreign news events coming from Greece and Portugal are only going to affect the Italian corporate risk through their effect on the Italian sovereign race. Under these three main assumptions, this Rigobon and SA type of approach uses these differences in the 2nd order moments in the volatilities in order to estimate my coefficient of interest, which in my case is this alpha 1. So in the paper, I do provide ample evidence that backs up this type of assumption. For example, I show that there are no important trade and financial links, and I also study what was the impact of this news on other European countries, on the volatility of the European stock market, meaning the market price of risk, and so on. I would like to skip the details of that because I want to emphasize the quantitative model that is coming next. So if you have any questions, I guess that I will be more than happy to talk about after the presentation, but if you don't mind, I would like to present my main results and then move on and discuss my quantitative model. OK. In terms of the results, what you observe here is this alpha 1 parameter or the estimation of that parameter. The first two columns shows both the IV estimate based on this Rigobonasack type of methodology and also the OLS estimate. The last two columns show the same set of results, but after you include some global controls, for example, changes in S&P, changes in the VIX index. Now, if you, you can see that all the estimates are positive and significant, of course. And if you take this value that is highlighted in blue, this is saying that a 1% point increase in the Italian sovereign risk leads to almost a 30 basis point increase in Italian corporal risk. In terms of economic magnitude, this estimate is accounting for almost 30% of the total increase in Italian corporate risk. So it's a very large number. Now this estimate is in the ballpark of previous empirical analysis, and my contribution to this literature is to show that behind this overall effect, there are very important heterogeneous effects across the non-financial firms, and that is the picture that you can observe here. So here you have the same IV coefficients based on this RigobonaA type of methodology. But for firms with different levels of risk. So the first dot and line that you have up here is the IV estimate and the confidence interval. When I include all the forms that I have in the sample. The next two cases are those in which I split my sample in 21 in which I only include the safe forms, and the other one in which I only include the risky firms, and the distinction there is the median level of pre-crisis risk. Now to the right of this red line, you have the same set of results that for the finer decomposition of firms. So firms here are sorted from the safest all the way to the riskiest, based on their pre-crisis levels of risk. Now, the key takeaway from this picture is that those firms that were safer before the crisis are almost unaffected by the increasing sovereign risk. Now, once you focus on this subset of riskier firms, those firms are significantly more affected whenever there is an increase in the default probability of the Italian government. So there is a very important heterogeneity across these financial firms, these non-financial firms, and that will be a central element, as you will see in the design of my quantitative model, and I will use these different reaction to increases in sovereign risk in order to provide some policy guidance on the optimal way to dampen the negative effects of this type of recession. OK. OK, so far, I have not said anything at all regarding the channels that may be behind this relationship between sovereign risk and corporate risk. So what I'm going to do now is to highlight at least the role of the banks, or what I call the bank lending channel in this transmission from sovereign to corporate risk. So in this very simple exercise, what I'm going to do is to use Italian bank level data. So I have data coming from the Italian Banking Association that covers the universe of all the Italian domestic banks. And what I'm going to do is to exploit the within region banks heterogeneity across sovereign exposures. So of course, this is not a causal estimate what I'm going to present here that is far from being causal estimate. It's just to highlight that those banks that are more exposed to sovereign bonds will affect or will restrict more their credit supply and that will affect more the non-financial firms. So that is the idea of this particular section. Controlling for this regional heterogeneity is important in the Italian case because the domestic banks in Italy are quite different depending on the region. So for example, banks in the south tend to be smaller, tend to be more concentrated or more exposed to sovereign risk, while banks in the north are typically the larger banks with a lower exposure to this type of risk. And what I'm going to do is to study or to quantify the differential impact on non-performing loans for banks with different levels of sovereign exposure. Let me skip the details of this particular section. Let me just mention the particular result that I have. And what I show in the paper is that those banks with a higher sovereign exposure, after controlling for this regional heterogeneity and a lot of bank characteristics such as the size, the leverage, the net worth, the reserves, and many more, they tend to exhibit a larger increase in their corporate non-performing loans, which is kind of like a good proxy for the corporate risk of their clients. The result that I have in particular says that if you take two banks that are one standard deviation apart in terms of their sovereign exposure, the bank with the higher exposure tends to exhibit an additional 15% point increase in the growth rate of its non-performing loans, and that estimate can account for almost 20% of the total increase in corporate non-performing loans that we observed during the recession. Now of course this is not a causal estimate. It is far from doing that, but I'm going to use this as a motivation for the quantitative model that I'm going to explain next. The idea is that these banks are an important driver in the transmission of sovereign risk to the non-financial firms. OK. Please let me know uh if there is a question. If not, I'm happy to continue with the quantitative model. No, there, there are no questions, but I'm gonna abuse my, uh, power of the chair, last one. So, so how are you distinguishing sort of the bank lending channel from, for instance, the fact that some firms may be more reliant on public procurement or government responders spending? That's the part I perhaps missed. Uh, yes, yes, I mean, behind this, this result that I show you right here, this is the overall effect. This is the total transmission going from sovereign to corporate risk. And as I said, behind this result there may be a lot of channels. The bank lending channel is one, the one that you're mentioning is another one. A fiscal channel may also be important or a trade channel may also be at play. Now the only thing that I'm highlighting here is in this very simple exercise that suggests that the band lending channel can explain at least for a significant part of the total transmission of sovereign risk. I'm not saying that this is the only channel that matters. In fact, there are more, but I'm going to take this as a motivation for the model that I'm going to present next. Understood. Thank you. Sure. OK. OK, so, so sorry, yes, yes, of course. So we, we can think that the, that when banks take uh exposures, sovereign exposure and who they lend to is not necessarily completely exhausted, completely independent. I can imagine that. If I'm, uh, I can imagine the local government telling, OK, you're going to buy not only you get a lot of certain banks, but my friend here and there, you're going to lend them money for their pet projects too. So, so that will create a lot of correlation between sovereign exposure and corporate. Of course, that is a, that is a fair comment. Uh, of course, I'm not going to control for all that. That's why this is far from being a causal estimate, but there is a whole literature that basically analyze the type of question that you have in mind. So this literature, what they're going to do is to use the type of Kawaa M and type of within firm regression. So they're going to consider one firm that is linked to two or more banks. And they're going to study the changes in the credit to that firm for banks with different levels of exposure. So in that way you can perfectly control for firm level characteristics, and what they show is that those banks with higher exposure tend to decrease even more their supply of credit to that particular firm. So that of course it is a causal estimate in that sense and it's controlling for all the concerns that you may have between the sorting across banks and the clients of that particular bank. And that is of course consistent with the result that I'm showing you here today. Al, please go ahead and ask your question. Uh, yes, Matthias, uh, just very quickly, so I know you were saying you know you see spreads just simply goes, I mean, naturally, those will be more like financial firms that have credit default swaps, and you want to have as large as a sample as, as possible with the publicly traded companies. But I mean, just to get, I, I know you don't want to push a lot on the, on the methodology of the outcome, but Uh, when you have looked at, uh, like individual like directive, um, you know, the, the implicit risk pricing for these firms like looking at, you know, options and Because my sense is that you're saying, OK, safer firms are not affected. Uh, riskier firms are when there's these 29 events, but a lot can be just because of uncertainty, and then that might be. So let me tell you what, what I did, uh, to control for all those concerns that is more regarding common factors, for example, a change in the market price of risk that may affect different, uh, the safe and the, the risky firms. So what I do as a check is to replicate the same analysis that I just did for Italy but for Germany. So I'm going to control or compare a firm, let's say in Italy with a 3% default probability with a German firm with the same default probability. And basically they are controlling for these differences in the profile of risk across the different firms. And what I show in that analysis that I have is that These foreign news events that I captured based on Greece and Portugal are only significantly affecting the corporate risk of the Italian firm but not the corporate risk of the German firm. So that at least provides evidence that goes in line with the assumption that there, that there wasn't a big change in the market price of risk across Europe because of these foreign news events coming from Greece and Portugal. Thank you. It's Sergio and after that, Carmen. Um, Maths, uh, I understand why you're analyzing, uh, listed firms, but by analyzing listed firms you're also constraining your analysis to safer firms. So you're looking at riskier firm within the safe firms. Uh, if you were to expand this analysis to all Italian firms, you might see much larger effects. And I was wondering why you haven't analyzed that if you have access to the Italian data, and I understand that measuring risk for those firms. It will be harder, but the transmission channel might be much larger. I completely agree with you. All these estimates, because I'm focusing on this subsample of firms that are large and publicly traded, can be interpreted as a lower bound in that sense because these are the firms that 1 may think that are going to be less affected by this sovereign shock because they have other sources of financing or they have more capital, more capital buffers, and so on. Now, I'm focusing on those farms because I need the daily frequency of the data in order to identify the shots. So even if I have access to let's say the universe of Italian firms based basically a small and medium enterprises, even in that case I won't be able to use this type of identification in order to estimate the pass through going from sovereign to the corporate sector. The type of identification would be different in that case, and the one that I'm using certainly won't, won't do the trick there. Carmen Uh, yes. Um, now, if you're a bank and you care about the overall risk of your, not just loan portfolio but your overall portfolio, for better or worse, you know, uh, we do view bonds often as the risk-free. Uh, uh, assets. So, you know, ex ante, uh, is it not plausible that the banks that have a higher share of, have more exposure to, to sovereigns. Uh, were willing to make riskier loans, uh, in the first place. Uh, so, you know, and when the sovereign risk was going on and this is going up, this is also the period in which economic activity, you know, was, was contracting. You, you, you know, you were in recession. So indeed, you know, you'd see higher NPLs, uh, but, but, but that, that. The overall portfolio risk decision. Yes. Uh, you know, what was, I do have, I do have an answer for that, Carmen. Thank you. Uh, of course that is something that I won't be able to show because I don't have that type of data. Now, the literature shows that this different sorting in the portfolio of the bank or these firm level characteristics are not driving the results. So this literature that I was saying. To Alvaro before in which they perfectly control for firm level characteristics using the same firm that is linked with two different banks. That literature shows that the firm level characteristics of the firm are not necessarily driven the result that they document in that sense. So even though this result cannot say anything regarding that because I don't observe that type of data, I don't have that granularity in the, in the data set that I'm using, I'm confident that these results are not driven by that sorting that you, that you are thinking because there are many papers that show that that sorting was not really problematic, at least in the Italian case. Please continue. OK. OK, thank you. OK. Now, let me jump directly to my coin identity model because as I said at the very beginning, this is the main contribution of my paper to this sovereign that type of literature. This is a brief picture of the model that I'm going to consider. So we have 4 different sectors, of course, the government, the domestic banks, the households, we are just going to be savers, and the corporate sector which is formed by heterogeneous firms. Now you can think of this as a closed economy, OK? And the key ingredients or elements are the following. So first of all, this is going to be a model in which both the government debt and also the corporate sector debt is risky because they can default. In that sense, this is a model with limited commitment. Now, as I said at the beginning, I'm going to assume for simplicity that sovereign risk is completely exogenous, so I'm not going to consider feedback effects between the real economy and the riskiness of the government. In this model, banks are going to supply the funds both to the government. They are holding the government bonds and also they're going to provide loans to the non-financial firms. And I'm going to consider some financial frictions, in particular, a leverage constraint. And because of that leverage constraint in which banks are going to be subject to, the model will feature the thing that corporate risk is endogenously linked to sovereign risk through the banking sector. And this is the idea. There are 2 different mechanisms in this model through which sovereign risk will end up affecting the non-financial firms. I do have an endogenous channel which I call the bank lending channel, and I also have an exogenous type of mechanism. Now the indulgence mechanism is the following. If there is an increase in sovereign risk, because banks of course are heavily exposed to that type of risk, the net worth of the bank will decline, and because of the leverage constraint, banks' lending capacity will be affected. Banks will reduce the amount of loans to the private sector, or they will charge higher rates. In that case, the firms have more incentive to default because now the interest rate is higher. They find it more difficult to roll over their existing stock of debt, and there will be an increase in the default rates and there will be an increase in corporate risk. On top of that channel, which is completely endogenous in the model, there is also an exogenous type of mechanism through which changes in sovereign risk directly affect the corporate sector. Although that is going to be in a reduced form way in my model, this exogenous channel can capture all the other channels that you may think other than the bank lending channel, for example, a fiscal channel, a trade channel, and so on, and inserted channel if you wish, and so on. And I'm going to use this extra layer of flexibility in the model in order to target the estimates that I presented in my empirical analysis, and I'm going to come back to that point. Now, the key ingredient that I introduced to this literature is this two-way feedback loop between the network of the bank and the corporate sector. So the goal of this paper is to study once we have this increasing sovereign risk, how the presence of corporate risk interacts with the financial system as a whole with the balance sheet of the banks, and how that interaction can significantly amplify the effects that were originated by the increase in the riskiness of the government. So that is the model in a nutshell, let me start describing the different sectors of this economy. So then let's start with the heterogeneous firms. Firms are going to be different across three dimensions capital, the long term loans, and also in terms of their productivity. OK? What they're going to do is to use their own stock of capital K. They're going to hire labor in the market L in order to produce the unique final good of this economy using this codalas type of technology. Notice that this technology depends on two different productivity processes. It first depends on this idiosyncratic productivity process which follows an AR1, OK? And that process is exactly the same across all the, the firms that I have in the model. But on top of that idiosyncratic component, it also depends on this same variable, which is an aggregate variable, and the idea is that this will allow me to capture aggregate productivity losses in the event that the government defaults. So the idea is that this sign will take a value of 1. It is normalized to 1 if the government is not in a default today, but it will take a value of something smaller than 1 if the government is in a default today. And I'm going to use this type of approach, of course, a reduced form way in order to exogenously link corporate risk to sovereign risk, OK? And I'm going to calibrate this particular variable in order to match the IV estimates that I presented in the empirical analysis. Just coming in with a super quick, super quick question, uh, why it so in, in the case of Italy we never observe it take the value of one, but in Greece we do, uh, why not. I, I, I know you're Italian, but, but, but why not include elements or, you know, even in a reduced way, uh, uh, the Greek. Uh, you know, test this to bring data where you do get, uh, you know, more action there. Yeah. Thank you, Carmen. That is an excellent comment. I'm, I'm from Argentina. I'm not from Italy, but I know that my last name may bring some confusion in that regard. But let me tell you this, uh, I'm going to use this value to calibrate, to match the IV estimates, but this value that I'm using as part of the calibration of the of my model is quite related. To a lot of papers in the literature that documents the drop in TFP upon a sovereign default. So for example, there is a paper by Guidos and Leis and coauthors for Argentina in which they estimate this value to be around 15%. That is the drop in TFP upon a sovereign default, and that is a very similar value to the one that I'm going to use in order to match the IV estimates of the empirical analysis. OK, thank you. OK, please let me know if there is uh another question here. Um, all right. So let me describe the recursive problem of these firms. So this is going to be the evaluation of the firm, and it will depend on the idiosyncratic states, capital, loans, productivity, and on the aggregate states which are captured with this capital S vector. So the owner of this firm is going to choose the optimal next period stock of loans, the optimal next period stock of capital, in order to maximize the current dividends plus the expected continuation value of this particular firm. And as I said, this is a model with limited commitment, so the firm may default. It has the ability to default on their loans. And that default decision will depend essentially on an outside option of the firm which is captured by this red term or this VD function, OK? So the idea is that to the extent that the continuation value is smaller than this exit outside option, the firm will default on its loans and it will exit the industry forever and it will be replaced by a new entrant. Now, this is of course going to be subject to a couple of constraints. The first one is nothing more but the definition of dividends. So in this case, the dividend of this firm is going to be the sum of the current profits pi minus the investment cost and there are a couple of frictions behind this function that let me skip. Plus the proceeds from issuing new loans. So this queue is an endogenous subject in the model and it is the unit price of each loan. And this depends on two different things, OK? It depends on the riskiness of the firm, meaning the default probability, and it also depends on the balance sheet of the bank. The idea is that to the extent that the banks are constrained today, this price will be smaller, meaning that the spreads will be higher, and that will have an implication on the dividends of the firm, on the evaluation of the firm, and therefore on the incentive to default. The terms in bracket is just whatever this firm is issuing, OK? In terms of new loans. So this MF parameter is capturing the share of loans that are maturing today. And the last term that you have here are the debt services, OK, so each period the firm has to repay some coupons, and also the share of loans that are maturing. The last two are just, uh, this is imposing that dividends cannot be negative, that is purely for simplicity. And the last one is just the perceived law of motion for all the aggregates of this economy. The household side of the problem is going to be very simple and straightforward, so I'm going to skip the details. The idea is that these agents are risk neutral. They can consume the unique final good of the economy. They can also save. They only have one instrument available, and those are short term deposits. They will receive some lump sum transfer from the government, and they are also the owners of the bank of this economy. Let me skip that recursive problem and let me focus now on the government side of the economy. So this is a model in which I'm not going to consider the optimal default decision of the goal. In that sense, that would be exogenous. So this is what I'm going to do. Each period, the government will issue some long-term bonds in order to satisfy its budget constraint. And that budget constraint depends on the profits of all the firms, but it also depends on some exogenous component basically that follows a fiscal rule, OK? And the idea is that the price of these bonds, it is also an endogenous object, QBS. And that object depends on two things on the sovereign risk of this government, meaning on the probability of default, and it also depends on the balance sheet of the bank or in the bank's stochastic discount fund. Now, how am I going to model the default decision of the government which is exogenous? This is following the Booda 2016 paper, and the idea is that in each period there is going to be a realization of this HG variable which can take two values, 0 or 1. If it is zero, it means that the government is not going to default today. If it is 1, it means that the government will default today. OK? And this realization depends on two additional things. It depends on some epsilon shock, which is an IID type of shock. But it also depends on the level of this S variable, and I'm going to interpret that as as sovereign risk. And the idea is that the larger the value of S, then the more likely that the government will default today for any given realization of this epsilon shock. And this sovereign risk will follow this very simple A1 process, and that is the only source of aggregate uncertainty that I have in the model. OK, to close this economy, I need to define the problem regarding the banks. OK, so remember, the banks are owned by the households, and the idea is that they're going to maximize their respective net worth a upon exit. So W is the value of the bank and what the banker is going to choose is the optimal amount of government bonds that they want to hold, the optimal amount of loans that they're going to give to each of the firms in the model. And the optimal amount of deposits that they're going to take from the households, OK? And what they're going to do is to maximize their expected net worth upon exit, which is this expression that you have right here. There are a couple of constraints, of course. The first one is just the definition of the balance sheet. So on the left hand side what you have are the sum of the deposits of the bank plus the net worth of the bank. On the right hand side you have the market value of all the assets of this bank. So the integral is the market value of the loans to all the firms. This last term is the market value of the government bonds that are being held by this particular bank. Now this constraint, which I call the incentive constraint or also leverage constraint, is the key equation that I have in the model. OK. This equation is saying that a fraction kappa of the assets of the bank cannot be larger than the value of the bank w. And why is this equation so important in my model? Because this will allow me to link endogenously sovereign risk to corporate risk, and this is the intuition behind it. Suppose that there is an increase in sovereign risk. Because the banks are heavily exposed to that type of risk, there will be a reduction in the net worth of the bank and therefore in the evaluation of the bank, so this W will decrease. Now, because of this incentive constraint, to the extent that this inequality is binding, the bank will have to react by decreasing the loans to the corporate sector. So there will be a reduction in the credit supply or there will be an increase in the spreads that they charge. And that of course will affect the incentive to default of the firms and therefore the level of corporate risk. The last two equations are just law of motions for next period net worth and all the aggregates of the economy. Let me skip this. Let me jump down to the results that I have in this model. So here we have some results for the non-stochastic steady state. So this is an economy without sovereign risk. And what you have here is the firm distribution both by size, that is the left panel, and also by default risk, OK? So the bars are for the model. The lines are for the data and as you can see the model. That's a good job in matching the two distributions. This is somewhat somehow calibrated, OK? But this is completely untargeted, OK? So it is remarkable that the model can capture this very asymmetric distribution in terms of the default risk of the non-financial firms, OK? And I will come back to that point later on. OK. As a source of validation of the model, what I'm going to ask is how well can this model replicate the Italian crisis, OK? To address that question, what I'm going to do is to fit the model with a sequence of sovereign shock, OK, that resembles the one that you observed in the Italian context, OK. The other three panels are endogenous variables in the model, and as you can see, the model does a good job in matching the size and the persistence of the crisis. OK? So the red line is the Italian data, the black line is the model in client dynamics. Now, it also does a relatively good job in matching the increase in the leverage of the bank during the recession and the posterior decrease in banks leverage once the economy starts to recover. And lastly, even though this is sort of a targeted in the calibration, the model does a good job in matching the dynamics of corporate risk during the recession, at least as implied by my empirical estimates. Yes, there's a question from Tuan. Yes, yeah, so, so there's something I didn't get. So from your incentive constraint, the kappa thing. Yes, wouldn't the model generate multiple equibrium because if enough firms default. Then a lot of firm will not be refinanced and will default. We strategically default, yeah, so won't you get intrinsically multiple sensotic equilibria in this case, yeah, I mean that type of concern, uh, is going to be present whenever the pricing kernel, sorry, this QB. is determined after the firm's decision regarding how much debt they want to take or after whenever they want to default or not. In order to break that type of multiply equilibra concern, this is the typical assumption that this model of default they have is that first the firm is going to choose whether to default or not after you observe the default decision. The price kernel of this function will be determined based on whatever you want to, to borrow from the bank. And that type of assumption allows you to break this type of multiplicity of equilibrium. Uh, is that OK, OK, perfect. OK, right, sorry, sorry, there's one more question from Paolo. Yes. Yes, so, in the, in the context of the Eurozone crisis, um, Italy was, was thought to be too large to fail. So what happened to the real economy in Italy and to the debt crisis in Italy would also determine the institutional decisions at the EU level. How, how, how, how can you think about that in the, in the, in the model here? OK, so basically that will be related to one of the policy analysis that I'm going to consider. So because there was this concern regarding these banks are too big to fail, I'm, I'm going to consider some policies in which the Italian government was injecting capital towards these banks, and I'm going to analyze in the context of my model what are the benefits of that particular policy. Of course, this model won't be able to say anything at all regarding moral hazard because those types of policies, they do have an important type of moral hazard component, but I will be able to quantify the effects of that particular policy in the context of the Italian economy. Sure. Yeah, I mean, in the bank um lending channel that you're considering, you're looking at the asset side of the bank, um, but the liability side of the bank can also be hit. Um, uh, there is some evidence that the banks usually have a higher withdrawal of deposits. When you have more links to sovereign risk, so that contraction in the deposit base also affects the lending capacity of the bank. That seems like you're kind of ruling out that channel. Yes, uh, for sure, I, I do consider that channel to be at play, at least in this very large Italian crisis. Uh, as you said, uh, this model is not going to be able to say anything regarding that. Of course, the type of policy implication that you may have in a model subject to bank ranks may be a little bit different to the ones that I'm going to present next because in that case you have more incentive to recapitalize the banks in order to avoid a potential bank run, but that is something that I'm not considering so far, but obviously it is very important to add that for future research in one of my papers. OK. Go ahead. Uh, I think one very brief observation is, you know, the, the, the narrative your, your, your model gives us is grossly oversimplified, but it's a common creditor channel in which you, you have contagion from the sovereign to the, to the, uh, corporates through the common creditor in this case, the banks, you know, there is an analogous literature on cross-border contagion also, uh, connected to the common creditor, but you know, it, it, it, it really, I, I mean, and it's simplest stripped down version. I think the common predator story is something perhaps you could, you know, um. Emphasize more. I see, I see. I do agree with you, and my contribution to that broader type of literature is to show the importance of the heterogeneity in the non-financial firms, meaning in the borrowers, and that is something that I think that it will be clear with this picture. So what you have here are the IV estimates for of the empirical analysis that is in black, OK? And you have the model implied estimates and based on simulations of the model. Now, as I said, I'm using this reduced form type of productivity losses in order to target this particular estimate, OK? This IV estimate. This is a targeted moment in my calibration. All the other IV estimates are completely untargeted in the mall. So the model does a really good job in matching this heterogeneous reaction to increases in sovereign risk. The idea, as I said at the beginning, is that the safer firms are almost unaffected by the increase in sovereign risk, but the riskier firms are significantly more affected. So in terms of the effects of the corporate sector in the financial system, The point that I want to make with this is that there is going to be an amplification mechanism going from the increasing corporate risk to the non-final to the financial system. But that amplification mechanism is driven purely by these very risky firms. OK. The majority of the firms, they are not going to observe a very large increase in their corporate risk after a sovereign shock. So those firms are not going to amplify all the effects. The problem is coming from this subset of riskier firms which are going to default more, and that will have an important impact on the balance sheet of the bank and in the economy as a whole. OK. So I do have 10 more minutes. So let me just summarize in 2 more minutes the decomposition of all the effects that are at play in the model. OK? So what you have here is the impulse response to a 3 standard deviation increase in sovereign spreads, OK? And you have 6 different endogenous variables in the model GDP, bank's net worth, capital spreads, and so on. The line in black is capturing the full effects in my model, OK? When all the channels are active. Now, in the first counterfactual that I'm going to do. is to analyze the effects that are coming from the contraction in the supply of credit only due to the increasing sovereign risk, OK? What I'm going to show is that under this counterfactual, You can explain only a small fraction of the crisis, but that is consistent with the previous literature. So, sorry. So here you have this counterfactual. This is the one in red. So as I said, this is a world in which corporate risk is almost constant and all the effects are coming from the increasing sovereign risk and how that increase affected the supply of credit. You can see that under that scenario you can only explain about 20% or even less of the total drop of GDP, so only a small fraction of the crisis. Now what happens when you move to a world in which you consider corporate risk? I will show you that you get a much larger drop in output. Of course behind this larger drop there are two things going on. You have the exogenous mechanism, OK, but you also have the indulgence mechanism which is the bank lending channel. Let me show you that under this war with corporate risk, my endogence mechanism, the bank lending channel, can account for 40% of the decline, and from this 40%, about half of this is explained by corporate risk. OK? So this is the control factor that I'm going to consider, the one in blue. So this is a world in which changes in sovereign or corporate risk are not going to affect the network of the bank. So the network is constant in the simulations. So all the effects that you observe in this blue line in this second panel are completely exogenous in the model and are coming from this reduction in firm's productivity, OK, so the difference between my baseline model, which is the black line, and this other control factor in blue, so this shaded blue area. This is capturing the share uh explained by my indulgence channel or by my bank lending channel. So this is about 40% of the crisis. Behind that decline, there are two things going on. You have the effect of sovereign risk on the balance sheet of the bank, but at the same time you have the effect of corporate risk on the balance sheet of the bank. So to understand the importance of corporate risk, this last counterfactual highlights the additional drop that you get in GDP once you factor in how this increasing corporate risk affects the balance sheet of the bank. So this gray area that you have here is saying that at the peak of the crisis, about half of this mechanism is explained by corporate risk. And as the economy recovers, the fraction is almost 100%, OK? And this type of effect was completely missing from this sovereign debt literature. And the main takeaway from the model is that You have a sovereignty crisis, the financial crisis, but you still need to carefully consider what is happening with the non-financial firms because these firms are going to significantly amplify whatever is happening with GDP through their effects on corporate risk on the balance sheet of the bank. And that's why I said at the very beginning that corporate risk can significantly amplify both the size but also the persistence of a sovereign debt crisis. And that is the new result that I'm bringing to this literature. OK. So, let me skip this. Um, let me, in the last 6 or 7 minutes that I have described the different fiscal policies that I'm going to consider, OK? So because I won't be able to analyze how changes in the fiscal deficit have an impact on the sovereign risk, this is the type of exercise that I will do. I'm going to consider 4 different fiscal policies, with the same fiscal cost, and I'm going to see which of these policies delivers a larger reduction in the drop of GDP during the crisis. So the first two policies are policies in which you are helping directly the corporate sector. Policy number one is one in which you are basically giving a debt relief to all the firms in the economy. Policy number 2 is one in which you are going to give this debt relief but only to the riskier firms, meaning firms with a default probability higher than a certain threshold. And the last two policies are more in line to what we observed during the last European debt crisis. For example, policy number 3 is the one in which the government is repaying all the non-performing loans. Policy number 4 is the one in which the government is injecting capital directly into the financial system. OK? What I'm going to show is that the debt relief program towards the riskier firms. Displays very important efficiency gains, OK? And let me give you the intuition and then I will show you the results. The intuition is the following. When you're helping these riskier firms, you can achieve two different goals at the very same time. On the one hand, Because of course you are helping those firms closer to the default boundary, you are going to obtain a very large drop in the number of corporate defaults. OK? That is pretty obvious. But now at the same time by doing this, you are decreasing the overall level of corporate risk in the economy and because banks are exposed to that type of risk, this type of policy allows you to indirectly recapitalize the banks and that of course is going to benefit all the firms in the economy. OK? So let me give you the results in the last 5 or 4 minutes that I have. So this is the fiscal cost for all these 4 policies and the policies that I want you to focus on is the one that is in blue. That is the debt relief towards these riskiest firms, OK? So in terms of capital or also in terms of the change in GDP, notice that this policy, the one in which you are helping the riskier firms, allows for a smaller contraction in GDP or in capital during the recession, OK? And the the difference seems to be significant according with respect to the other three policies. So why is this policy better? The intuition is the one that I was giving you before. Notice that the change in the default rate under this policy is significantly smaller than the increase that you observed in the other three policies or under the baseline and scenario grade without any policy, OK? And the idea is very intuitive. You are helping the firms that need it the most, OK? That's very obvious. Now the key point that I want to emphasize with this paper, this is the key ingredient that I want to convey is that this policy which you are helping the non-financial firms. Can have a very positive impact on the network of the bank, OK? So that is something that you can observe in this last panel. So the policy in which we are helping the riskier firms is the 2nd to last bar that you observe here. And notice that the reduction in the network of the bank is very similar to a policy in which you are directly injecting capital towards the bank. So the amount of the reduction in the network is very similar in the sense that you're going to get all the benefits from this policy, but at the same time, you are saving this policy, these firms from a default, which is a good thing. Now, given that you are able to recapitalize the banks, you are going to observe some positive spillovers towards firms with small or medium levels of risk which are not directly targeted by this type of intervention. And that is something that heterogeneity is something that you can absorb in this last figure that I have for today. So the first panel shows the change in the default rate for firms with different levels of risk, safe, medium, and risky firms. The bottom panel shows changes in the capital stock of these firms for safe, medium, and risky firms. So this policy in blue is targeting only a subset of the risky firms, OK? Mostly half of these firms are being targeted by this intervention. And of course, by doing this, there is going to be a large reduction in the default rate and also a large contraction in the drop of capital that is kind of obvious. But the key idea is that once you do this, you are also recapitalizing the banks and that will create a benefit on firms that are not directly targeted and that is something that you can Observed in these two panels right here, so the contraction in capital is going to be dampened for the medium and for the safe firms because given that the banks are more capitalized, they will be able to have access to cheaper credit or to a larger credit supply. So I know that I'm running out of time, so just let me conclude. In the empirical analysis, I use Italy purely as a case of a study during the last European crisis using an identification strategy that I exploited the volatility of Italian sovereign risk around some events. I showed that sovereign risk can account for almost 30% of the total increase in corporate default risk and also, and this is my contribution to that literature, that there are very important asymmetries across France behind that transmission. Then I provide a very simple type of exercise in which I show that the band lending channel seems to be an important driver behind the transmission and that is consistent with a very large empirical literature. Then I took these two facts to motivate a quantitative model in which I showed that corporate risk significantly amplified both the size and the persistence of the crisis, and the idea is that this increase in corporate risk will further affect the bank's balance sheets and that will contract even more the credit supply. Lastly, consistent with my Pican analysis, I showed that the riskier firms are more affected by increases in sovereign risk. So I studied different policies that exploit that heterogeneity in order to provide some policy recommendation, and what I show is that those policies that are geared toward the riskier firms are more effective. Because they present positive spillovers through the interactions with the financial system and those positive spillovers end up benefiting the safer firms of them all. As the next step is to introduce some macroprudential policies in this context to assess the benefit of those policies. Thank you very much, everybody.

showAllTranscripts

no

duration

PT1H15M51S

scene7File

worldbank/MatiasMorettiSovereignRisk

scene7Domain

scene7FileAvs

worldbank/MatiasMorettiSovereignRisk-AVS

title

MatiasMorettiSovereignRisk

description

MatiasMorettiSovereignRisk

showTimestampAndTranscript

yes