 So the title of my talk is private debt booms and the real economy do the benefits outweigh the costs So this is sort of going to be focused more on the macroeconomic implications of credit booms and not to leave you in suspense The short of my take is probably not Because it requires quite a bit of humility to kind of answer this question Both because I'm not going to be taking you know a stance on the welfare implications Our credit booms which are very very difficult to do and also because even from a statistical perspective This is a this is a hard question to answer But I am going to argue that from a just a macroeconomic perspective in terms of you know income per capita real GDP Credit booms don't really seem to be worth it in the sense that the benefits don't seem to outweigh the costs Okay, so let me define what is a credit boom or what is a private debt boom Well, I'm just going to think of it as an episode of a rapid expansion in credit originated to household and non-financial firms and we're really interested in private debt Oops because they've been playing an increasingly prominent role in economic fluctuations in business cycles At least in the post war era, especially since the late 1970s credit booms have kind of been mattering More and more for booms and recessions The way I think about it from a really broad perspective is there's two ways to think about Private credit or private debt in the economy and about credit boom specifically One perspective is kind of the more traditional perspective at least before the crisis Which is the idea of credit deepening that is that there's a credit boom could represent a period of structural improvement In the financial sector's ability to intermediate credit to households and especially to firms That leads to more investment that should be good And that's sort of if you basically read a lot of the literature up until 2007 That was most of sort of the finance and macros Discussion was about credit deepening In recent years, we've become much more attuned again to the cost of credit booms Which I'll summarize sort of with the term financial fragility this idea that credit booms lead to builds up in risk and Vulnerabilities that end up sowing the seeds of economic destruction in different ways that they'll be more more specific about So the basic question in some sense about credit booms is whether the benefits From the perspective of credit deepening outweigh the costs From the press that come from financial fragility. So what am I going to do in this paper? I'm essentially going to revisit the connection between Private debt booms and the real economy that's been the subject of a lot of research by many people in this in this room And looking at the evidence in a number of different ways both just what are the consequences of credit booms? For for growth and also how do credit booms actually affect the real economy? So what are the mechanisms that we should be thinking about Perhaps credit affects the real economy and the way I'm going to do that is I'm going to be using new panel data Not put together by myself But put together by a lot of hard work by people at the BIS and the IMF Covering about a hundred and forty countries over the past six decades or so So the coverage depends a little bit from country to country and along the way I'll talk about how this connects to all of the all of the fantastic work That's been done by many people in in this room So the key message number one is we really have to think about credit booms and credit deepening as fundamentally different Different economic mechanisms they operate in very different different ways And I think this provides some bridge between the literature on kind of finance and growth And the literature on credit booms and financial financial crises So in the data long-run growth does go hand-in-hand with credit I think this is actually very important to To remember credit is a kind of a good barometer for the level of development both within countries Over time and just across countries in levels and you hear this if you talk to policy makers Central bankers, especially in emerging markets the level of credit is something that they monitor as essentially a good sign of the economy However in contrast to sort of this more gradual credit evening debt booms are actually associated with short-run economic booms followed by predictable and And the predictability here is reasonably strong economic slowdowns in terms of real GDP in in in the medium-run and One of the things that you see in the data, especially in such a large sample Is that I think there's reasonable evidence that they also not only lead to slower growth in the medium-run But actually to a lower long-run level about and this is I think more controversial statement So let me just be clear about that and the reason why credit booms and credit deepening are sort of different And we have to think about them differently is that they affect the economy through fundamentally different different channels So debt booms there's lots of evidence I'll show you distort the economy essentially by boosting demand in the economy instead of productive capacity And by fueling unproductive things like real estate Booms that turns out to be a systematic pattern up in the data and once the booms and they leave they leave a lot of Problems and trouble for for the economy and for regulators So banking sector distress debt overhang in the private sector and also for small open economies They leave an overvalued a real exchange rate and this is both small open economies outside of the US But also within the US across US states you see that areas that have the biggest credit booms actually end up losing competitiveness Okay, so let me just talk a little bit about this idea of credit deepening because it is in some sense an idea That's been forgotten a little bit in the past few years I I think at least in terms of the emphasis in the research So there's a large literature that argue is just that a better functioning financial system Contributes to and facilitates a stronger economic growth kind of you know, perhaps the seminal Kind of summary of this is this handbook chapter by Ross Levine And this this sort of research generally argues That credit deepening improves the economy through some sort of productive capacity channel typically So a better functioning financial system lowers the cost of credit for firms to invest it increases access to Constra to credit for constrained firms It maybe leads to a better allocation of credit across firms So less milk misallocation Across firms and lots of other different channels that are good. So in this literature credit depth measures Are typically used to make proxy for financial So a way to think about it is well private debt booms. They might just be periods of accelerating credit So here's the evidence on credit deepening This is just a very basic data on the level of private debt to GDP in 2015 and log real GDP per capita in 2015 as well I call this the income debt curve, but there's probably a you could probably come up with a better name But it's actually kind of a pretty you know strong positive relationship in in in levels So there's no structural differences across countries like what I would do that But you know, it's strongly positive and to some extent a calm cave. So maybe suggests that It's not as positive for more developed more developed economies But you tend to see that you know countries that have more developed financial systems tend to have higher income per capita A lot of this is obviously running from you know, just financial development To credit to GDP, but maybe no that's at least what this Financing growth literature argues. Maybe there's some you know causal effect from credit to a higher level So this turns out so this is just a level across countries if you look In the cross section of countries over time What you see is also that countries that grow more over time in terms of real GDP also have more rapid Increases in credit to GDP, but the relationship is not nearly as strong as as in level So this is just sort of a long-run change over over 40 years So it does suggest that sort of long-run economic growth does go hand in hand with with credit And that and that might be partly because of you know, the influence of better functioning financial markets That you know allow firms to access Access credit Okay, so now let me kind of zoom in with that backdrop and talk a little bit more about Private private debt So the question is really our private debt booms just periods of accelerating credit Interestingly, I don't think many people in this room would think that but if you go present in a typical macro seminar And you show that credit booms predict lower growth They would say that's impossible permanent income hypothesis or some other type of logic, you know Credit should follow good things, you know in part because people should have the right beliefs and and so on and so forth So what I'm going to do to kind of Try to answer this question. It's just use a very very simple approach. That's Inspired by work by Richter Schorich and Wachtel, which is just try to identify a bunch of credit boom Events so here I'm going to identify a hundred ninety credit boom events for a hundred thirty three countries going back to around 1960 All of the results I'm going to show are really robust to using more traditional PR sort of, you know, Georgia local protection type methodology as well But I think this is just kind of a very simple way to show the results So this is how I identify a credit boom just to use the US as an example What you see here is just the level of private debt to GDP In the United States going back to 1960 What I do is I use this filtering procedure. It's called the Hamilton filter But you can think of it like an HP filter or just some kind of methodology that tries to extract Some sort of long-term trend. So it turns out the Hamilton filter, which is Based on a paper by James Hamilton called why you should never use the HP filter So I kind of thought I should use the Hamilton filter It turns out it's a little bit kind of more weakly up and down But it does kind of do the job of detrending So if you take the difference between these two figures What you get is this Hamilton filter private debt to GDP ratio for the United States and it sort of makes sense We have two major credit booms since 1960 in the US. We had the 1980s a credit boom Which was kind of somewhat more moderate and then we obviously have the 2000s Okay, then to actually identify a credit boom event. You need to do something very arbitrary You need to special specify some threshold. So I'm gonna specify a threshold that is every time Basically this Hamilton filter debt to GDP exceeds 1.64 times standard standard deviation Right, so so whenever kind of this Hamilton filter debt to GDP is above some value And you can sort of vary this value and think about robustness So what this implies for the US is that there's one debt boom In since 1960 which which starts in 2005 and then a debt boom is just identified as being in the first year of Whenever the series crosses this threshold Okay, so then let's just ask using this very simple approach What are the dynamics of you know real GDP around these debt boom events? So I'm gonna look at Real GDP growth from five years before and then you know as far out as essentially I want to go so I'll go Ten years out. So basically the dating of the private debt boom is in in time zero So this would be like 2005 for the US boom. So I'm gonna start in 2000 And then basically just gonna ask, okay, what happens to the predicted path of real GDP? So if I knew how to make a gift and put it into a Presentation that's what I would do, but this is gonna be kind of the more simple version. So this is what happens So basically real GDP goes up up until the dating of the debt room So so far it sort of you know looks like things are going reasonably well But then what happens is after the debt boom is dated you see a drop in real GDP and basically the drop essentially continues For you know ten years out So this is consistent with kind of this evidence that shows that these debt booms tend to predict now at least in sample and to some extent out of sample slower growth In in the medium run that I think has been kind of established in a number a number of different papers The more crucial thing here is that there's even some level Some evidence that the long run level of output is actually lower than the previous trend Let me get back to that point and make sure that I Okay, so I think actually this evidence is not in some sense too surprising if you connect the dots Between existing research that's already been done. So for example Shuleric and Taylor and others show that debt booms predict financial crisis There's an older paper by Sarah and Saxena in the AER that shows that financial crisis actually leaves Permanent output gaps actually output basically falls and then you're always below trend. So you're never going to catch up Okay That doesn't necessarily mean that debt booms are going to lead to a lower long run level Output because it could be that the debt booms that don't need to crises actually end up leading to very strong growth But in work that I've done with a technique on our Sufi We found that debt booms and especially household debt booms, but debt booms in general predict gross slowdowns more generally So there's you know this predictability for financial crises is an offset by some predictability sort of in Okay, so that's how it connects with the existing evidence. Let me just mention a couple of caveats So one the long-run effect is really quite difficult to estimate for a very simple Statistical reason that if you're increasing the forecast horizon Just the uncertainty increases with with the horizon The second caveat is kind of one of causality So I've essentially just taken these debt boom events and sort of argued Well, let's just look at what happens to GDP and sometimes, you know, use the word association and maybe sometimes use the word That sounds more like causality So there's of course this issue that get rooms aren't actually just these exogenous events They could be caused by sort of some more fundamental economic factors, but that being said if you think about it Most of these most of the kind of Typical models that you would write down would argue that debt rooms should actually follow good economic developments or good good economic Conditions so it sort of at least does raise some important questions about debt rooms Why debt rooms tend to be associated with lower growth and maybe a lower long-run level So my conclusion is believe that the hypothesis that debt booms are associated with some Beneficial credit deepening seems to be quite strongly rejected by the data. So debt booms So sorry credit deepening is is potentially associated With with higher levels of income, but debt booms are just not the way that you're going to get that credit You think that booms are fundamentally fundamentally different You know, these results are kind of reasonably robust. You know, if you go back Look looking at a longer event window around these episodes are kind of going 10 years back and 15 years forward You know, you can look at other types of specifications like not actually doing any of this identification of these debt booms I'm just looking to let's look at three-year changes in debt to GDP and how they correlate with kind of GDP growth around these these periods and you tend to see sort of similar patterns You know, you can run kind of local projection type impulse responses and you find similar patterns Actually, local protections don't even give you much of a boom in the short run Okay, so let me talk now a little bit about This question of you know, why are debt booms associated with lower future? I'm gonna kind of try to dig into the mechanisms a little bit more and tell a little bit of a story about sort of the typical debt boom in this post-war In this post-war sample and I think some of the mechanisms or the story I'm gonna tell is gonna help reinforce At least to some extent this idea that debt booms are actually Causing lower output lower growth Not just sort of associated with some other economic economic mechanism Okay, so the first the first fact really is that debt booms are Seem to be associated with or maybe driven by credit supply Expansions earlier. We were having a discussion about whether a credit cycle was just a credit supply cycle or not And here is sort of evidence that these debt booms tend to be credit supply driven So you have private debt to GDP here detrended just around these debt boom events So now I'm going back and using these debt boom Events and here you have measures of credit spreads which you see tend to be low During the debt booms and then they spike at the end of the book So this is kind of very consistent with work for example by Arvin Krishnamurti at Tom and your who presented earlier that credits was tend to be low Kind of before during debt booms and then they spike So credit supply expansion that seems to be kind of the main story For for the for these debt rooms credit supply expansion can be driven by lots of different things It could be about beliefs could be about financial liberalization It could be about global liquidity. I think that's very important for lots of small open economies Just the cost of borrowing in international markets and when those costs change that leads to an increase in the availability of Okay, so basically the fact that credit booms are associated with an expansion in credit supply Already tells you explains why credit isn't just following for example faster productivity growth So there's something that's actually changing in the financial sector. That's leading financial intermediaries to be willing to lend more So when you have this credit supply expansion, it can basically operate yet in two ways One it can boost productive productive capacity or two it can boost it can fuel Demand and in a recent paper with Autisman and Amber Sufi We kind of write down a very simple open economy model that lets you disentangle the role of these different Factors, so in particular if the credit expansion is operating through a demand channel What you should tend to see is an expansion in the non-traded sector relative to the traded sector and what you should also see is an appreciation of the real exchange rate and for open economies where you have our first countries We have the data you should also see Rising rising imports as as well and the logic is very simple If you have an expansion in credit supply that increases demand for example by households That leads to an increase in for example consumption demand if you consume tradables and non-tradables Well, you can import more tradables, but you have to actually produce the non-tradables locally So that leads to a big reallocation toward non-tradables But non-tradables are going to be relatively scarce so that increases the price of non-tradables the price of non-tradable goes goes up That's effectively a real exchange rate Appreciation, so that's that should be happening if the demand channel is operating if the Productive capacity channel is operative in most models you tend to see limited Reallocation between the tradable and non-tradable sector even if the credit expansion is affecting these two sectors differently, and what you also should tend to see is rising rising exports Okay, so what do the data say again for these debt boom events across the countries? Well, what you tend to see is an expansion in non-tradable relative to tradable employment during the boom followed by a sharp reversal After after the debt boom the same was true for non-tradable relative to tradable output So this is evidence that these debt rooms seem to be affecting the sectoral composition in the economy kind of a little bit connected to what Carsten was saying earlier earlier today using sectoral credit data So they seem to be kind of leading to this reallocation toward non-tradable sectors like construction, but also Also, for example wholesale and retail trade at the same time they lead to real exchange rate Appreciation as well. So this is the real effective exchange rate from the BIS You can also think about real exchange rates as a relative price of non-tradables and these debt booms are associated with with real appreciation In both of these measures that means that it makes it more difficult for the tradable sector also to compete So that's sort of another consequence of these debt booms On the imports and exports and net exports margin You see imports are rising kind of relative to their trend and then collapse after the debt boom in Contrast exports are actually a little bit above trend But kind of actually seem to be dwindling and fall a little bit less So that means that actually during these debt booms these economies are importing lots more goods and running kind of increasing their current account Trade and current account deficits also borrowing more externally in order to finance finance these These reports that provides I think further evidence for the role of these kind of this kind of demand channel in Kind of how the debt booms propagate to to the real economy Okay So then you kind of you have this demand boom and then at some point We sort of know from the data that the credit supply expansion tends to reverse So there's some kind of a meaner version in credit market sentiment and I think one of the things and you're on and The previous paper by Alessia and Casper was sort of getting a little bit at this what actually precipitates the reversal I think this is one of the things that's much less well understood It's definitely less well understood by professional forecasters because they usually don't predict the reversal. So that's But I think we also don't understand the reversal that well either and at the same time so One sorry one kind of potential hypothesis is sort of this more behavioral story Of kind of a sequence of bad shocks that leads to a shift of from optimism to pessimism That shifts credit market sentiment and credit markets become much much much tighter Some evidence for this you know including working with Matt and Wade that Matt talked about yesterday Is that bank equity returns seem to suggest that relatively early in crises and kind of after credit Losses tend to be realized by bank equity Investors suggesting that maybe for there's a gradual realization that many of the loans that were made during these debt rooms Weren't actually as profitable as expected. So this year is bank equity returns during these debt boom events So they're kind of soaring but then they peak a year or two before and then they really fall and there's essentially a permanent decline in Market value of bank equity around these these episodes and it happens a little bit before When when the overall credit cycle versus in contrast non-financial equities seem to keep rising until about the peak of the debt Boom and then they fall and recover Someone again another kind of interesting asymmetry between bank equity and non-financial equity is kind of the more medium-run Implications that for banks the losses seem to be permanent So sort of like a cash flow effect that the end of these booms tend to be bad for banks was for not for non-financial The losses and more temporary suggesting that you know risk premium are relatively elevated during these And of course you also see these dynamics for house prices and if housing booms and busts are really a essential feature of lots of these debt booms especially the most destructive ones as for example the paper By Moritz and his co-operative show that actually one of the best variables for diagnosing whether a debt boom is bad Is actually the growth of house prices? Okay, so what are the factors behind? The growth slowdown. I've already talked about credit supply reversal that sort of triggers it You know the other factor is just all the debt that you know all the debt overhang all the debt You build up during the boom under a certain level of expectations that has to be paid back. And so for example a really nice Really nice paper by Anton Kornick and his co-author shows that the GDP declines after debt booms Coincide exactly when the net flow of funds reverses from lenders to borrowers so basically when borrowers have to start repaying When when when when when when lenders stop lending and when borrowers have to start repaying We kind of looked at this so with my co-author Giozogliannozzi We looked at this during a very kind of extreme episode Which was this foreign currency debt crisis in Hungary were essentially households had to you know repay after the debt boom But they had to repay about 60 to 70 percent more than they expected because most of the loans were Denominated in Swiss banks and there was a large appreciation of the Swiss Frank during The financial crisis and we see that this of course leads to large rise in defaults Big collapse in consumption and also very severe local recession in areas that have more of these with strength So sort of debt overhang itself seems to be an important important central mechanism, you know the decline in house prices and other asset values like you know Octatomia and Amar-Sufia have showed as well as many others that Kind of precipitates a worse worse recession as well and sort of reinforces the decline in demand And then what you really need You know Based on most models and also the evidence that we have what you really need for the decline in demand Translates into a decline in output is you need some notion of downward nominal rigidity For example downward nominal rate wager to do that actually translates those declines in demand into declines In output like for example the model of by Schmeckrohe and One of the interesting things about downward nominal wage wage rigidity or just nominal rigidity in general Is that they can also imply kind of more long-term? Effects so when you have these debt booms you have periods where you kind of you see there's overvaluations So real exchange rates appreciate and then if it's hard for wages to actually fall Then you can have kind of more long-term effects as the tradable sector loses Competitiveness and this can sort of lead to persistent misallocation And kind of a loss of market share amongst the tradable sector that can maybe explain part of the reason why these debt rooms Not only lead to gross slowdowns, but actually a lower long-run levels Okay So I think I'm close to being out of time So let me just kind of provide a few final final thoughts some of these are correlated with with Carsten's Thoughts as well. So what do we actually do about these these debt booms? So, you know macro potential policy Has has kind of received a lot of attention and I think they macro-approved policies do have a role in kind of limit limiting the most excessive the most Extreme debt booms, but I think we also have to be aware of the fact that they can't be expected to avoid all Crises and they have a few practical Practical challenges. So the first one is unintended consequences and we talked about these unintended consequences in a number of different ways This this conference so part of it is the migration problem. So you regulate one sector that leads to migration to other sectors You also have unintended consequences, even if you do regulation for example at the sector level or the product level that it leads banks to lend more to risk your borrowers in other Sectors so that we have direct evidence from from Ireland for example to you know, there's this time inconsistency issue Political cycles. So Carsten Muller has a nice paper showing that you know these macro-prudential policies are subject to to political pressure potentially and the third one which I think is actually kind of the most Fundamental challenge is in some sense. How do we really know when the macro-prudential policy actually worked? So for monetary policy, you know, we have an observable two percent target for inflation We can sort of observe whether we're at Whether we're at the target overshooting or undershooting. We have a full employment target as well That's perhaps a little bit harder to actually evaluate But still we can sort of we can take data to it For macro-prudential policy. What what is kind of the criteria? Is it that we should have no crises at all? Is it that the crises should be less severe less severe relative to what I think that's where? Conferences like these all the research that's being done just in terms of understandings how these credit cycles work And that's sort of how they influence the economy is very important precisely because it gives us a better sense of how these Policies would have worked and I think you know there probably is sort of the Stigler rules So George Stigler had this rule for flying that if you're not if you don't miss a few flights You're probably spending too much time in airports So if you don't have a few crises, maybe we're kind of over like over-regulating some part of the financial system And I think that is just something that we have to be aware of kind of as we think about Formulating all of these policies. So yeah, thanks a lot for the invitation and thank you very much for having me discuss the paper As expected, it's a very insightful paper Thought-provoking and I just have a few thoughts on this So Emile's thesis is squarely Based on on these pictures which Emile Explained so these pictures so that following a debt boom or credit boom, which is a rapid short-run expansion and private credit to GDP Real GDP first increases then decreases But the focus is on the long-run effects where in the long run it may end up at a lower level Than where it initially started so and and and this basically gives rise to to Emile's conclusion That you know the idea that debt booms are part of beneficial credit deepening is soundly rejected by the data Because it looks like in the end we're ending up at a lower level of GDP where we started so these debt Booms must be ultimately bad In real terms now before I talk about this Let me just quickly point out one one thing this result and these pictures are also Obtained in in Emile's paper with Amir Natip Which focuses more on on the role of household debt So you have the same impulse responses which show cycles and ultimately we're ending at a lower level Then where we started in that paper they they quickly mentioned that but they don't emphasize it for two reasons One is as Emile pointed out the standard errors become very large as you increase the forecast horizon another one however is they also notice that The lower long-run level of GDP after the positive shock is not robust After excluding the greater session that was in the case of the the Jota local projections We notice that so one thing this what you did in the QGE paper You looked at household debt and here you look at overall private credit to GDP But it may be worthwhile checking whether or not this is robust to include in the greater session or or not It's just a small remark so The main question I have is What is the counterfactual here in all of these pictures that we've seen and the results are obtained based on comparing? GDP relative to some trend or to some Country-level average because you use fixed effects in these regressions and and the question is is that the right counterfactual? If we did not have this debt boom, would we be worse off? Would we better off would be a GDP trend or country-level average or maybe is the counterfactual a different growth path all together and This may sound not intuitive Because we always think that when we end up at a lower level of GDP relative to trend that must be bad But there is a theory out there that would argue that these credit booms are inseparable from the beneficial effects of credit deep link and ultimately that there is a genuine trade-off between a long-run real GDP growth and Crisis risk, so there are effectively two types of growth path One is a risky growth path, which has high GDP average long-run growth But also credit booms and crisis risk and then there's a safe growth path Which has no credit booms, but lower long-run GDP growth and according to this theory, which I will explain in a little bit the Choosing a growth path a safer growth path without these debt booms would be like throwing out the baby with a bathwater Because we would not have the debt boom, but we would be worse off because we have a lower long-run average growth now the other View of the world, which I think Emil would more subscribe to is that these credit booms are in principle separable from the beneficial effects of credit deep link and you know Policy can do something about them to some extent at least Whether or not we can avoid them, but at least they can be crook their impact can be softened and in this case I believe The right discussion has to be about what can we do about it? Emil mentioned a little bit in the end of his presentation, but I want to talk a little bit more about that how can we identify these credit booms in real time and What are the policy tools available? I will talk about this in the context of a study that has not been mentioned so far in these two days It's a study by IMF researchers based on 170 countries Good so and when we look at the United States 2004 To 2009 and here you have private debt to GDP and the green line is real GDP per capita This kind of looks like the impulse responses that we've seen from From emails presentation and and and if you look very carefully we're ending up with a lower level of GDP then then we're started and and so this is like the The adverse long-run effect now when we take a bigger perspective and we zoom out we see that This is like what we have on the left side here We see that in the long run. There's a very positive correlation in private debt to GDP and real GDP per capita in The US and this alternative view of the world is let's call it the trade of theory would say well Maybe the counterfactual of not having this little wiggle would be having a lower long-run growth path So we can't really not have this wiggle here. I mean I'm calling the great recession a wiggle So we can't not have this wiggle here and and assume we would have continued along some trend in the absence of it but the alternative would have been a much lower lower growth path altogether and this view comes from a very influential paper by Russia Tornell and Westerman in the QGE and and they're motivating pictures are Thailand and India I'm not sure if you can see this at the end This is India the solid line and the dashed line is Thailand and he had real credit and he have GDP per capita and Ultimately India, which is the solid line here has a very slow but very steady growth pass through a hundred forty fourteen percent GDP growth Thailand on the other hand has a very high growth but they also have lending booms and bus and they have associated crisis, but the The bottom line of this picture is despite the lending boom and bust Thailand has a higher long-run GDP than India And so the argument is that obviously the authors do not want to say that you know financial crisis are good for growth But what they want to say is that you know high growth path are associated with the undertaking of systemic risk and with the occurrence of Occasional crisis the cost of a high growth path is this crisis risk and The theory behind this is Nice theory is based on imperfect enforceability of contracts which lead to boring constraints and in their theory model there's an implicit bailout guarantee which then leads agents in the economy to take risk systemically and this leads to investment and Lowering the firm's cost of capital because the lenders get repaid in the bad states due to the bailout now as a result you have more investment in growth in these periods without crisis But then the downside is you have also higher likelihood of crises because you you take systemic risk, but the overall effect Weighing of these costs and benefits is positive assuming contract enforceability is not too high So as long as there are these borrowing constraints that matter the overall effect is positive and ultimately the economy with With risk-taking has a higher long-run growth in the economy without risk-taking Empirically the authors I get to this using a measure of financial systemic risk Which is a skewness of credit growth and the idea behind the skewness is in a crisis You have a large and abrupt downward jump in credit growth and because crises are rare The distribution of credit growth will be negatively skewed and here you have India here You have Thailand and you see Thailand was the one with the boom bust episode in in lending and therefore you have you have a left skewed And distribution of credit growth here You see also consistent with the pictures earlier the mean growth of real G of real credit in Thailand It's much higher than in India Thailand was also more volatile than India and Thailand as I said is negatively skewed So in their empirical work, they basically try to understand what affects Growth what affects real GDP program with capital growth and what they find is that Obviously higher credit growth on average is good for GDP growth. This could be the credit deepening effect More volatile credit growth is bad for GDP growth. This is well known But now importantly skewness and affects credit growth in the sense that that we see in it earlier That it means the more left skewed credit growth is the higher is GDP growth meaning this crisis risk is positively associated with the long-run growth and And they actually do show that this skewness here captures it could capture a lot of things But they're careful to clean it up like so it does not capture wars or large terms of trade iterations Good. So this is one view of the world. It's admitted in extreme view in the sense that we can't really do much about Or we shouldn't do anything about The these step booms because the only alternative would be a little growth path now the other more policy friendly view and is that you know, maybe we can separate these credit booms from from credit deepening and This is a study. I want to discuss a little bit. It's by some IMF researchers based on 170 countries I'm sure many of the audience are aware of the study And they do something very similar to what a male is doing They define a boom episode as a deviation of credit to GDP from trend and Emile has one point six or times the standard deviation. They use one point five times the standard deviation So the idea is is is is very similar to find a hundred and seventy six boom episodes The median credit boom last three years during this time credit to GDP grows at 13 percent per year Which is about five times the growth in non-boom years Now in their study 32% of these booms are followed by a financial banking crisis within three years 62% are followed by below trend real GDP growth, which is similar. What Amy was showing and Are these higher or lower numbers? It depends on whether you want to see the glass half empty or half full and 30% of credit booms are not followed by either a bank banking crisis or low or long run GDP growth And So the first thing they ask is something very similar to what a man is interested in what is the long-run effects of a credit boom and I just read this out because they formulated pretty nicely It's a well episodes that sharply increase credit to GDP ratio have a long-term beneficial effect depends on two factors The first is the extent to which credit booms contribute to permanent financial deepening And this is this is a notion that Amy will say is Rejected or that you would not subscribe to and they conclude in about 40% of the episodes Pretty to GDP seems to shift permanently to a new higher equilibrium level In fact, there is a positive correlation with long-term financial deepening and Accumulated credit growth that occurred during boom episodes The second is then the extent to which financial deepening resulting from a sharp increase in credit is equivalent to deepening achieved through a slow gradual growth and Here they conclude that there is a positive correlation in the data between the number of years a country has undergone a credit boom and The cumulative real GDP per capita growth achieved since 1970 So this is consistent broadly with this trade-off view that there is this positive correlation between credit booms and associated crisis on the one hand, but also higher growth now The next question is if we think these credit booms are separable from credit deepening, what can we do about it? Should we do something about it and what can we do about it? And so they also talk about, you know, when do these credit booms happen and in about 30% of the cases They happen in association with financial liberalizations That actually whose purpose is to foster financial deepening, but then they ended up in a rapid credit growth Searching capital inflows after capital account liberalizations are another contributing factor and then very often strong economic growth It's a predictor of credit booms. So often credit booms follow strong economic growth, which is effectively demand-side story, I think And how do you predict these good versus bad credit booms in real time? It's not just about predicting credit booms in real time Because if policy wants to do something about it, you want to break the bad ones and separated from the good ones, right? And so here they don't have very much to say because it's really hard to predict this in real time They they conclude that larger and longer booms and those that start the higher level of personal GDP They're more likely to end up badly and asset prices are a Predictor of credit booms, but they're not a predictor that allow us a separate good from bad credit booms So they find that when they look at equity prices in general they grow at about 11 percent per year during these booms But the 11 percent is the same for good and bad booms So they don't allow us to separate the good and the bad booms. At least this is what these researchers conclude And when we talk about policy, I Kind of do jump in we could have a whole discussion about about policy here So let me just briefly do this on my last two slides and They look at monetary policy fiscal policy in micro crew and their conclusion is that For monetary policy the problem is many times credit booms are associated with very quiet macro conditions So then kind of monetary policy would do more harm than good Also in open economies obviously when you type of monetary policy and raise the rate Then you have capital inflows which will undo the effect of title monetary policy overall They find relatively little empirical evidence that monetary policy can lower the incidence of these credit booms fiscal policy is rather ineffective in their view Mainly because there's a significant time like and then there's political economy Considerations in the data. They find it actually has the wrong sign and this could be mainly reverse causality Lastly macro potential policies. It's probably according to many the most Serious policy tools that we have to tackle the most promising that we have to tackle credit booms. So And generally the idea is the one is smooth and financial credit cycles and To prevent systemic crises and in cushions against the adverse effects if the crisis occurs They're targeted unlike monetary a fist the policy which of course also makes them susceptible to political lobbying and Circumvention because the targeted policy. So the three and that they discuss is I mean one we've seen in an Oscar's Presentation earlier on capital requirements. They find actually the same result as Oscar Presented basically the capital requirements Have little success in reducing the incidence of these credit booms But when these credit booms occur, they're beneficial because you know, you have built up buffers that allow That allow banks to this is to weather these storms Better and then credit growth limits these are speed limits on the on the speed which with credit can grow and They have some success. They argue But the problem with like with many of these targeted tools is whenever you impose a regulation restrictions there are ways of getting around it and and as we all know, I mean when lending by banks was curtailed non banks shadow banks or whatever we want to call it came in and they filled the void and and and and so Just some success then then lastly LTV or that the income requirements They're potentially affected by those relative little empirical evidence. They come through So let me finish. It's a nice piece. Obviously. I expected no less from email In the text a fairly strong stand Maybe a little too strong the credit booms can be separate from credit deepening. So the alternative is maybe We have to choose between two different growth paths And but if that is true, then I think we want a little maybe know a little bit more about the implications for policy Like how can we predict these credit booms in real time? How can we separate good from bad booms in real time and then you know talk a little bit more about policy? I think one of the deeper points are raised by this runs here and Tornell work is sort of a broader question which is you can really choose between two economic systems One in which you don't have much credit deepening and you you're never gonna get credit booms But you're gonna be on a slower kind of growth path And you know if you can choose that kind of without liberalizing your your financial sector without opening up Capital account or you can choose kind of this other world in which you're gonna have some booms and busts But your trend is gonna be is gonna be higher. I think I I think that's a very very kind of You know Interesting and kind of provocative way to think about it I don't actually have an answer for whether that's true or not whether whether countries should choose one path or the other or Whether there's something in between I think I'm making us a kind of a slightly more modest claim Which is that within these credit booms that happen in these countries that are Liberalized these credit booms don't seem to operate in like the way that the Ron Sierra Tornell model or most models would predict They would affect the real economy Which is that sometimes you take on a lot of risks to invest in some projects that turn out not to be productive and because they're not Productive and you have some type of financial accelerated and that leads to a crash. These are really other types of booms They're more kind of demand type booms and in that sense In that sense, I think of them as kind of different different animals are operating through Different channels and I think that also does provide a little bit of help for policymakers in real time in understanding Good booms versus bad booms essentially what I'm saying is that most of these booms on average There are more bad booms than good booms But of course there are some good booms as well And one way to separate them is to look at these different measures like the real exchange rate like the non-tradable relative to tradable employment and output share in order to see whether you're having these these kind of Distortive effects that are leading to reallocation Toward the non-traded sector toward construction that tend to be more likely to be associated with us like for example Carson's work showed but also this will work by colonists that shows that no financial crises tend to be preceded for example My increase is a non-tradable relative to tradable type So I think that's you know, that's kind of a great broader question That I'm actually thinking about and struggling a little bit with myself in terms of what broader regulatory regime Do you want to begin? So, I mean and just leave that a little bit and then happy to take questions Of those so-called response functions because if I take them very literally the way you presented them So there is a credit boom at time zero, which causes GDP growth to be positive at t-minus five Yeah, so that you know, that's that's when it's dated. So that's the p right I mean another I think maybe more sensible way to think about this is that there are some economic expansions You know, some of them, you know may have a lot of expansion in credit and some of them may not You know and the ones that have a lot of expansion in credit lead to more severe recession Which we already know from the work of roga from Reinhardt and many other people I Really have a bit of an objection just from a purely Econometric and also just you know commonsense macro perspective as sort of thinking that a credit boom at time P is causing GDP growth at times I just just need to help me with this term because I don't see it and And so and then the other thing how robust is this to your definition of credit boom because of course There's this threshold and you haven't had time to show us, but I'm just wondering, you know How that would hold up to different definitions. It's going to be strange Yeah, so two questions There the first one is kind of just very basic in terms of the timing time zero is just when you date the peak No, no, I have to show you look at the average over time It starts about five years before maybe a better way to do it Which is kind of what we did in our QJ paper was kind of looking at you know structural BAR is trying to find episodes of Increases in credit supply that lead to credit moves like periods when credit spreads tend to be low For example, and then tracing kind of the full boom and bust in credit in real GDP to to kind of these These underlying shocks instead of you know defining these events In in this way, you're right also on the point of you have underlying positive productivity shocks do also lead to credit expansion But these episodes actually are often much more gradual in these very rapid credit booms But I think just in the data seem to be much more driven by this broad umbrella term that you know I'm calling and many others are calling credit supply, which is sort of an increased willingness to mend kind of forgiven characteristics of risk that you see for example in credits read measures in credit composition measures As well, so that's kind of why I think about those as different from sort of periods When you have kind of good economic developments that lead to kind of more gradual credit growth But on the credit composition side, we have at this point some very mixed evidence particularly for the most recent boom I don't doubt the fact. I mean, I'm not questioning the evidence of spread So in that sense, I mean robustness to how you define it Sure, yeah, no, yeah Absolutely, and I mean, I think there's lots of different ways to do it Very these specials you can estimate AR's you can look at look at these things in other ways And I think most people who looked at these data can define the same thing Especially with the growth slowdowns with the longer run level point. That's a that's a more contentious First of all, I think hoger makes an interesting point and could have used China instead of Thailand Yes, it's the you know most profound example that I would have to do my picture myself. So I just copied the thing So I would make a couple of comments I think the heart and the soul of a crisis is as much the over capacity being created The physical over capacity being created as it is the credit It's too many buildings are being built or too many houses or too many railroads and then in the 1800s And to that extent, you know with a couple of the Presentations including yours had the question what precipitates reversal Well, what precipitates reserve reversal is the realization that there aren't enough tenants for the building and Then what precipitates the lower growth after the fact is we can't build any more buildings or houses for five or ten years Until demands catches up with supply The one last comment I'll make you know Your comment about credit deepening is an important comment And you know it strikes me as profound that absent calamity debt private debt, but overall debt Always outgrows GDP it always outgrows GDP and folks cite that as you pointed out as evidence of a maturing and Progressing economy. I think that's true to a point. I think it's a two-edged sword I think there's a point at which that's too high But I think you can't observe a single country or society where that growth does not continue So to me that's I think the dilemma of our age is that debt to GDP will continue to grow and it has a stultify a Regressive effect after it hits a certain point On the first point just about over capacity. I think yeah That's definitely one of the factors that leads to kind of losses on On these loans that precipitates the reversal and there's lots of work that sort of tries to separate for example The role of over capacity which has one kind of very clear policy implication that you should basically go destroy a bunch of houses or railroads from kind of you know having too much debt or having an under-capitalized banking sector which has quite different Policy implications, and I think I would actually say over capacity is a possible impossible to achieve without that So do you take into account fiscal policy different types of fiscal policy after the boom because I mean I guess that in the past Governments tried to balance the budget, which is per cyclical and in other cases it was expansionary So that obviously is an independent variable that has an effect of growth that we wonder if you've looked at that considered Yeah, no, that's that's a good question I think it's come up actually two or three times in this country And so Holder mentioned the paper the IMF paper that showed that it's a more fiscal space than you know The government can actually cushion the crisis more the Romer so Roman Romer have a recent paper looking at the aftermath of Banking crises and they argue if you have more fiscal space and you can smooth out the crisis more you can deal with your banks You can deal with your debt problems. You can stimulate. I think that that makes makes a lot of sense And we can look at that or I can look at that in this context as well Yeah, but preserving fiscal space during good times is I think there's a lot of evidence that that helps Yeah