 So let me start with a big picture. Inequality in the US and in Europe, but especially in the US, has been rising since at least the 1970s. For example, nowadays the top 0.1% of the population in the US is making as much as the bottom 50%. You can explain the national increase in inequality by looking within regions. Most of the rising inequality can be attributed to regional inequality rather than between regional inequality. Understanding how inequality can potentially be a factor in financial stability is an important question. And in particular how regional inequality can affect financial stability. So this is our research question. Is it true that regions that are more unequal in the US also have more unstable banking systems? And by unstable banking system I mean whether banks in those unequal regions are also more likely to fail. So a methodological approach has two parts. First we are going to examine statistically whether there is a link between income inequality on the one side and bank risk on the other. And then if we do find such a link we are going to ask what is the mechanism? And we are going to construct a theoretical model with which we are going to try to rationalize why do we see this pattern in the data. So first I'm going to start with the empirical part. We need three essential data sources. First we need data on income. Second we need to define what we mean by region. And third we have to define what we mean by bank risk. First let me start with income. We are going to use the US Census Bureau for data on income. Since 2005 they conduct an exhaustive survey called the American Community Survey where they essentially ask people about many different things including their sources of income. For our purposes this is the most valuable source of information because it contains its first voice representative and it's very detailed so it allows us to study inequality at a sub-national level in different regions. And second of all it contains data about poor people as well. Those that are usually not paying taxes. So we can study the entire distribution of income and not only those that are making more money. Next we need a way to measure income inequality. And for that we are going to use the genicoefficient. A genicoefficient essentially an intuitive way to think about it is if you pick two people from the population what is the absolute difference in income? A genio of zero means that everyone has the same income. A genio of one means that extreme inequality. One person has all the income. Next we have to define what we mean by a region. And for that we are going to use something which is called a metropolitan statistical area. So these are areas which contain at least one core city and they are selected so that they have a high degree of socioeconomic integration. So for example the biggest metropolitan statistical area in the US is Newark Newark and Jersey City which has almost 20 million people. The smallest one has a little bit more than 50,000. There are almost 400 metropolitan statistical areas in the US. For our purposes what is important is that the banking regulator uses these metropolitan statistical areas to define banking markets. Next we have to have a measure of bank risk and for that we are going to do two things. First we are going to look into our sample and see how many banks in a given region failed during the period of our sample. And second we are going to use statistical models to predict with what probability a bank is going to fail or what is called the predicted probability of default and that's going to be our second measure. And finally we are not interested in the risk of any individual bank. We want to talk about the health of the banking market in a given region. For that reason we are going to aggregate the bank risk. So I'm going to be talking about what is the fraction of banks that failed from 2005 to 2020 in a given metropolitan statistical area. What is the average bank risk of the banks in a given metropolitan statistical area in the way we're going to have a profile of bank risk that is ascribed to a given MSA. And we're going to see how this profile relates to the income inequality in that same MSA. First let me start with our empirical findings. What we find is that regions with higher income inequality have a larger fraction of failed banks. So yes, more unequal regions are also more unstable. What we also find and what is surprising for us initially is that not all banks in more unstable regions become more likely to fail. Some banks become safer while others become riskier. The challenge is to explain why do we observe this pattern and for that we are going to develop a theoretical model that is going to generate these two predictions that more banks are going to fail but also some banks are going to take more risk and become more likely to fail while others are going to take less risk. We're going to link income inequality on the one hand to bank instability on the other through the mortgage market. So we're going to assume that banks are going to choose the level of risk that they are taking. Here the crucial part is that the incentive structure in the banking sector. So since deposits are insured banks are tempted to take more risk. So that's the starting point of our argument. But at the same time we are going to impose the equilibrium condition the way we call it in economics is that at some point in time all banks have the option to start taking more risk. So some banks are going to choose to become safer. Other banks are going to choose to become riskier but at the time when they are selecting whether to take more risk or not they are going to be indifferent. Otherwise if they are not indifferent then all banks are going to choose whether to start taking risk or not and then one cannot clear the market. So for example if all banks want to take risk then no one is going to lend to the same borrowers and vice versa. What are our model implications? Well our model has two main implications. First within a region some banks are going to choose to become safe and other banks are going to choose to become risky. Second if you are a risky borrower in which in our case is a borrower with a lower income you are going to strictly prefer to borrow from risky banks why this is because they are going to offer you an easy credit and they do that because risky banks are competing among each other to attract risky borrowers. Now we can see how one can link this pattern to the distribution of income. If your region which has a high income inequality you are going to have a relatively large subprime mortgage sector which means that there will be a large fraction of risky banks and so as a result of that in these regions you are going to see a large fraction of banks that are going to fail. Second you are still going to have safe banks which are going to be necessary to satisfy the demand for safe credit. So overall the model can rationalize what we've seen in the data. So with regards to relevance I would like to highlight four points. First there is this discussion about what are the consequences of income inequality and with respect to researchers have examined many different consequences. Generally the message of this literature is that income inequality creates undesirable socioeconomic consequences and can destabilize political outcomes. What our paper is doing it adds to this list. It shows that in addition income inequality can also destabilize the banking markets. The second point of relevance is with respect to the data discussion which is about how is income inequality related to financial instability and there are different mechanisms which have been proposed in the literature and what our paper is doing is highlighting an additional channel. This channel has a particular regional component which has not been explored to a great extent in the literature. The third point of relevance that I would like to highlight is with respect to the subprime mortgage market. It's been widely accepted that in the run-up to the subprime mortgage crisis in 2008 there were a lot of toxic mortgages. For example there is something called the ninja mortgage which is no income, no job and no assets and there have been different explanations which were put forward to explain why did banks issue those mortgages. For example, securitization incentives. They were just issuing the mortgage and they're selling it to a different institution or misguided expectations. They were so optimistic about the ability of the borrowers to repay that they were willing to share a mortgage like this. Our model provides an additional rationalization which is from the model perspective these banks were essentially reshifting so they were not trying to securitize the mortgage. They were not having misguided expectations. So we provide a complementary channel which can explain why we were observing these sort of very toxic and high-risk mortgages. And finally the final point of relevance that I would like to highlight is with respect to the policy implication. What should the regulator do in that case? And first if banks that are in unequal regions are more unstable so one implication would be that the regulator should pay more attention and impose stricter regulation. Unfortunately though by imposing stricter regulation on banks the regulator is also going to hurt low-income borrowers who are benefiting from easy credit. And so as a result of that we're pessimistic that this sort of regulation will be able to be implemented because it has undesirable redistributional impact. This is what might make it possible for this reshifting to take place in the first place and for the regulator to turn a blind eye to it. There are two potential avenues of future research. So now the way we think about inequality just like many other authors is that we look at overall inequality and we're not looking at why is it that we have more inequality. Inequality can increase because of the declining manufacturing sector. It could increase because of a skew-based technological change. The decline of union power and business-friendly regulation and other factors as well. Perhaps not all inequality is born equal. Maybe if you have high inequality because the manufacturing sector is declining then that will have even more detrimental impact on the banking sector. So this is something that we are not looking in the current paper but we believe this could be an important future research topic. Second is the question is about polarization. Now we are looking at inequality and we are not distinguishing between what is the reason for this inequality. But another just as important aspect is polarization which is rich people living next to other rich people and poor people living next to other poor people. And our hypothesis is that polarization could be just as harmful if not more harmful for the banking sector than inequality. And the fact is that even a small change in inequality through polarization could exacerbate the impact on the banking sector. So this is something that we are also not looking in the current paper but we believe that it is important and could potentially provide interesting insights.