Understanding US Housing Bubble and Subprime Mortgage/Financial crisis

My friends keep asking me about the US mortgage crisis and if something similar is happening in India. Thought I would put some of these thoughts in a blog and here it is. Sincere thanks to everyone who pushed me to write this…..

PS: Incidentally, just as I write this, former Federal Reserve Chairman Alan Greenspan has presented a paper to the Brookings institute on Mar 19. In that paper he blames the rise of China and the dollar funds from China searching for a safe haven in US mortgage bonds, thus lowering the mortgage rates and increasing loan availability. Further the same money disappeared causing ‘destabilizing debt problems’. Adjustable rate mortgages with low teaser rates in the first few years and higher rates later are blamed too. Adding capital requirements is considered too. However, in my opinion, assessing the risk and sufficient capital requirements for a given loan considering the macro environment in interaction was a challenge. More about this in the modeling section.


Most people think that US mortgage crisis was just caused because of loans given to subprime (low FICO) customers, but that is not the case. There were several factors that contributed to the bubble and the resulting mortgage crisis. In fact, so called prime or high FICO customers being overleveraged was a big contributor to it.

2003 and 2004 with low interest rates – US was barely out of the dot-com led recession. Mortgage rates were at all time low bcos banks had lot of excess cash and Fed wanted to stimulate the economy which had just come out of the downturn. Federal fund rates were hovering around 1% until the start of 2004. Home ownership was being touted as one of the biggest ways to retirement planning – an asset that never falls in value, saves you taxes as well as gets you rid of rent. Construction industry buoyed a lot of other industries.

Investment banks came out with a study based on data that the low FICO customers do not have any greater loss rates compared to the high FICO ones based on the historical data. Semi-government rating agencies like Moody’s, Fitch and S&P also agreed with the investment banks by agreeing to rate a large part of these loans to low FICO customers as AAA meaning very low risk. One of the key things these guys overlooked was that historically low FICO customers were not given home loans and the very fact that there were some low FICO customers in the vintage meant that there was something positive about them which encouraged the banks to give a loan to them.

2005 selling the dream of home ownership – The era of ARM and Stated Income loans – Economy was upbeat, interest rates were still low. Everyone and his dog in California was buying a new, bigger home and a bigger gas guzzler! At million $ prices!! Before you even know about open houses, they were sold. You are sitting there and thinking – wow what did I do wrong in my life that I don’t have that sort of money; only to realize later that all these were the famous ARM (Adjustable rate mortgages) which had very low interest for 2 or 3 years and were pegged to Libor after that. The ideal plan was to milk the bank and get out of the loan before the ARM expires – the mortgage agents laid out the entire plan for that.

They even had stated income loans in case you did not want to give income verification or if your income was irregular due to your business nature. Can you think of getting a loan of half a million without proof of income. Yes, bcos you had the house as collateral anyways; and although a 95% or 100% of the home value has been given as a loan, the price of the house is going to increase anyways…… so it does not matter for the bank (or the mortgage agent) if you had an income or not.

2006 Innovative mortgage products and securitization put home ownership and prices in overdrive – In 2006, the same thing continued. However the competition between banks increased. At the same time the demand for high yielding junk bonds – read risky mortgage loan paper increased. The investment banks and rating agencies were right(?) – these loans were not risky at all and paid high interest rates. Securitization market boomed. Simply bcos when the customer could not afford the high interest rates or as soon as some equity was created in the home price – Not by the virtue of his paying back any loan, but bcos the value of the home increased – he would either sell the house and move to a bigger one or refinance it with another loan from another bank. Speculation was ripe with people buying multiple homes and flipping them. Meanwhile the investment banks showed huge profits on the junk bonds and more demand got generated. Banks showed large profits on Option ARM loans with interest accrued but not collected yet. Luminaries like Warren Buffett had started sounding alert by this time. But the regulators were busy with markets like credit cards where they prohibited loans that were negative amortization – meaning balances where the balance keeps on increasing due to customers not paying even the interest that is due. But wait a minute, these were typically loans under $ 10 K. How about those ‘Interest only’ or ‘Option arm’ loans which allow you to pay just the interest or even LESS than the interest. Isn’t negative amortization restriction more necessary for a $400k loan than a $10k loan? But mortgage industry was the sacred one unlike the evil credit card industry and all those home builders increasing the prices were still doing a great service to Americans by fulfilling their dreams of big home ownership.

Mortgage Banks’ focus on Wall street - Senior executives are excessively focused on Wall Street performance, not simply bcos it determines their bonus but bcos they are addicted to that glowing investor face. Metrics like Net interest margin and average FICO are looked at by the Wall Street to assess profitability and risk. Hence banks would do everything in their means to maintain these numbers. What it meant was to have a highly profitable portfolio while admitting high FICO customers. Metrics like FICO do not account for the loan taking ability as they do not consider income. They simply consider past behavior. This gives a false sense of low risk when in fact the individual may be highly leveraged for his salary. Very few people had time or willingness to pay attention to these factors.

2007 The first glance – Feb 2007 to be precise – HSBC bank gets rid of Bobby Mehta in a fallout of mortgage loans where the losses are much higher than expected, attributed to an error in the way expected losses were modeled. Turns out later that HSBC was the better one of the pack. Countrywide and WaMu actually hit the street with much higher delinquencies end of 2007. Part of which were being suppressed from the earlier numbers simply bcos they were booking much higher balances at a much higher pace. Countrywide in fact had a TV ad running which said that you can get not just a mortgage when you refinance but also get a low cost cash advance of $25k without any proof of income!!! This easy cash was financing a lot of consumer spending and home makeovers that people thought were capital assets they were generating, realizing little that there is no value for their resale.

2007 last quarter and 2008 – The denial and the happening – The senior management of all the banks were still in strong denial in most of 2007 and somewhere in between there was a Fed mediated buyout of Countrywide by BofA (this wasn’t made public initially). The rest of the banking industry was still in denial and even received great bonus for the stellar performance in 2007. The Fed started putting out the gravity of the situation and there were huge upticks in the mortgage delinquencies with the ARM expiring, interest rates at all time high and resale market for homes just vanishing in thin air. As consumer spending lost the source (ie home equity loans), economy was expected to contract, companies stopped hiring and started laying off employees which aggravated the fear further. To add to it, the crude shock was there once again and the psyche of people was hit. There was a cascading effect of negativity with people clinging on to their jobs and not buying any major items or making any big investments. Oh and the biggest thing – the home builders were still adding new inventory of homes bcos they were still making a profit by doing so. And yes, there were bank owned homes to be had for a bargain too. Home owners were stuck in the spiraling effect of denial and at the same time reducing prices – the pain of selling below the buying price and seeing all your equity evaporate in no time is too much. But if I cannot pay the increased mortgage after the ARM has expired and cannot refinance the loan bcos there is no equity left either bcos of current drop in prices or bcos I never had any in the first place, the only option was to sell at a loss or walk away. If the customer walks away, the bank is stuck with the property and the only option for them is to sell at a loss or be stuck with it.

Marked to Market portfolio – Typically, banks would get rid of foreclosed homes promptly. But in this case, as banks portfolios were marked to market. Every time they sold a home at a lower price, they had to provision for all the homes that they were holding on their portfolio in that market. Hence they did not want to sell at a lower price and there were no buyers at the higher price. At the same time, as this fear gripped the investment banks and their investors, they stopped securitizing mortgages. Not just the junk bonds market disappeared in thin air, but even the prime bonds market disappeared almost totally. So the only way for banks to issue mortgages was to keep it on their books. This was not a way most of these banks were setup as their mortgage engines were just processing houses which packaged the loans and sold them to investment banks. With no buyer in view, they practically stopped issuing mortgages or put fanatic terms like 35-40% down payment. As a result, even the willing buyers were unable to put up that kind of money to buy a house and the glut of houses with those due to moving, job losses, arm loans and lastly but not the least new constructions by home builders kept on increasing. In fact, several localities newly developed with the mortgage boom became ghost towns. All of a sudden all the optimism about home prices never fall seemed to disappear and everyone was gripped in fear.

Equity infusion – 2008 saw most banks like WaMu and Citi scrambling for equity to cover their losses for the provisioning. What several investors misjudged was the effect of systemic glut of houses, accounting rules on capital, unemployment and above all the effect of a bank losing customer confidence.

Loss of depositor confidence - As US deposits are secured only upto an amount of $ 100k a lot of large depositors started withdrawing their deposits any time there was loss of confidence in a bank. Although they were few in number, they had large amounts in the banks for whom every penny mattered. This led to further need for cash and lowered the capital in place. This probably was the last nail in the coffin for most banks.

Unemployment - Corporations started downsizing to account for lower business or at times just expecting a lower business or higher productivity given that workers are unlikely to leave. Outsourcing was increased at the same time with India and several other countries doing the plum jobs.

Fed steps in –Finally the Fed steps in to bail out banks with a stimulus package and prints more dollars. Typically no country other than US (maybe UK) is capable of doing this. As the currency is not pegged to gold, any amount of dollars can be printed as long as the economic sanctity is maintained. Other countries which hold huge chunks of dollars denominated paper have no option but to maintain the currency value in the short term atleast – this is the China’s dilemma.

Learnings from a Modeling perspective- Most of the banks in this crisis were using the best data and sophisticated models. So what went wrong?

The big components of any modeling are understanding the source of data, the assumptions and how the model is going to be applied (the limitations) and then interpreting the results in the light of those limitations and assumptions making sure that they are always borne in mind while making decisions based on that model. The mortgage models were build with data of sober home prices and rational behavior. If someone with a low FICO was offered a loan in the past it was with great caution and because of lot of other factors like maybe the deposit he has with the bank or long term relationship or an NGO backing it up. There were very few flippers in the high FICO customers. Having about 10 to 20% own equity was a must. Model development in big banks is sometimes so disconnected from the actual business that the bias of these factors is seldom captured in the models. The modeler at times is not even aware of it or ignores them as they form a very small percentage of the overall population. However when the same model gets extrapolated or the mix of the portfolio changes in a manner that a large portion comprises of different type of individuals, there is a risk to be considered.

Assumptions such as high FICO customers are low risk, home prices will always keep on increasing, homes can be resold to pay the loans, loans can be refinanced easily, incomes will grow at 15% enabling the customer to pay higher installments after ARM expires, unemployment never goes above a certain percentage need to be understood well. Simply because someone (read investors and investment banks) is buying junk mortgage bonds does not mean they are worth that amount. Market is not the most efficient way to price – it can very well misjudge the risk or just play along to make a buck. Transferring risk via derivatives does not work if the exposure of the other party is way too high leading to a counterparty credit risk.

The misjudgment on the cumulative effect of easy loan availability, increasing housing inventory, ARM resets, unemployment, increase in market share at the cost of risk, in the name of innovative products, and later denial of sharing the bank equity by infusing new capital at an earlier time were the bane of the mortgage industry. From the modeling perspective, incomplete understanding or half hearted implementation of the assumptions and limitations for the population and its interaction with the macro environment contributed to this mess.

Is something similar happening in India….?