Housing Bubbles And Their Financing

Housing finance is interesting, and offers an interesting take on some theoretical issues. Although the theoretical issues sound abstract, they are critical issues in economies facing a housing bubble. This article looks at one aspect of housing finance: the limit to financing is credit risk, not funding. Monetary flows in a credit-based economy are circular.

Note: I hope to follow this up with one or two article discussing the Canadian housing bubble. This article covers some basic points, and I will get to the more hair-raising topics later. I have broken the discussion up as the full discussion would be too lengthy.

One question that often arises in the discussion of debt markets: where will the money come from (for a particular borrower)? This is the wrong question. For a variety of reasons, our culture has created a mythology around the concept of financial saving. There is a widespread belief that we first need savers to make loans to borrowers. If all transactions involved the transfer of precious metal, this would be the case. However, like most commercial societies, we use credit instruments in most cases during transactions. In a credit-based system, flows are largely circular.

The figure at the beginning of the article shows the financial flows in the typical purchase of an existing house. We will assume that the home buyer is a new entrant to the housing market, and borrows almost the entire purchase amount (which is unfortunately typical in Canada at present). The home seller is exiting the housing market (moving into an old folks home?).

Money flows in a loop around the system.

• The financial sector lends the money to the home buyer, in exchange for a mortgage.
• The home buyer passes the money over the seller, in exchange for the deed to the house.
• The home seller will return the proceeds to the financial sector, in exchange for financial claims of some sort. (The home seller would first get a claim on the banking sector, but that could be immediately transferred to various financial funds.)

In popular discussion, there is a fascination with the banking system. In fact, the formal banking system may only be a temporary enabler of the mortgage lending process, with all instruments off the banking system's balance sheet within days. Therefore, economics 101 discussion of bank multipliers and so forth is spectacularly misleading.

• Even if the mortgage is made by a bank (and not a specialty mortgage lender), the mortgage will often by securitised. In Canada, the Canadian Mortgage and Housing Corporation (CMHC) -- a full faith and credit Federal Crown corporation -- guarantees mortgages for which insurance is required (discussed below), and these mortgages are pooled into Canada Mortgage Bonds (CMB's). In the United States, the housing agencies have been involved in securitising mortgages for decades (a fact which economics 101 textbooks have not yet caught up with).
• The proceeds of the home sale are unlikely to remain as deposits; they will instead be withdrawn and recirculated into the financial markets.

Since we cannot be sure what the seller will do with the proceeds, there is no single story which captures how the circular flows work out. In practice, all we can see are the aggregate flows, and note that the flows all (roughly) net out.

(There is a strange popular fascination with currency -- for example, dollar bills. One could try to tell a story where the home seller takes the proceeds in currency, breaking the financing loop. The reality is that most law-abiding citizens do not do this; nobody sensible wants to be known to be walking out of an office with hundreds of thousands of dollars in cash. My reading of the situation in Vancouver is that the opposite is the case: currency is being returned to the system via money being laundered to purchase homes (or at least, it was). In such cases, the buyer would presumably want to keep a low profile, and not take out a mortgage. I am not familiar enough with the financing practices of Vancouver drug gangs to make definitive statements.)

What stops the circular flow from going to infinity is credit risk. The flows will only take place if the borrower is seen as an acceptable credit risk by whatever segment of the financial market is underwriting mortgages. (Note that "acceptable" credit risk may take into account the credit spread versus default risk.)

In Canada, the determination of credit risk has been largely handed over to the bureaucrats at the CMHC (it has some private sector competitors). If you have less than a 20% down payment, you are required to get mortgage insurance,  if you meet the administrative requirements and roll the mortgage insurance into your mortgage payments, the CMHC guarantees the mortgage, and the financing flows.

(In the United States, private sub-prime lenders will compare the perceived default risk versus the credit spread. The Financial Crisis highlighted the weaknesses of this credit risk assessment. Things have hopefully improved since then.)

In summary, we have the answer to the perennial question: where will the money come from? This question is usually raised in the context of governmental borrowing (which perhaps tells us about the political biases of the questioner). The answer is that the borrower will generally transfer the proceeds to some entity -- that will then plow the proceeds back into the financial sector. There will be a flurry of portfolio rebalancing, but the flows will net out. (I discuss this again in the final section.)

Does Increasing Debt Cause Home Price Increases, Or Vice-Versa?

One area of debate within economics is the causality between mortgage debt and house prices. The role of private debt is a source of discontent for heterodox economists, with Steve Keen being a major voice. My opinion is that debt levels reflect what is happening in the economy.

The figure above shows Canadian household mortgage debt (only) as a percentage of GDP since 1990. (Total household debt provides scarier numbers.) One may note the hockey stick change in trajectory that started after the 2000 recession. There was a similar hockey stick that developed in home prices at about the same time. (My writing laptop has been acting up on me, so I was not able to get the house price chart done in time. The reader will have to trust me on this, or find another data source for the chart.) One could look at other countries and see the exact same pattern.

From a micro perspective, there has to be a linkage. We know that in practice, households largely borrow most of the payment for a home (that is above the equity value of the home that they may be selling). Down payment norms are set as a percentage of the home value, and that many households stick to the minimum payment. If home prices rise by 10%, the required borrowing will also rise by 10% (assuming an unchanged down payment percentage). Therefore, we know that mortgages outstanding have to rise.

(As a technical note, since I am showing the amounts as a percentage of nominal GDP, what matters is the growth rate versus GDP.)

Therefore, if we argued that house prices follow some fixed trajectory, this will cause debt levels to rise -- matching the observed data.

You could try arguing the causality in the other direction, but you run into a technical issue. The mortgage debt does not appear until the transaction takes place -- and the price is set as part of the transaction. The timing is incorrect.

In Canada (and presumably elsewhere), banks will pre-approve mortgage amounts to households. These pre-approvals form some kind of contingent instrument on bank balance sheets (a commitment to accept a liability?), but I am unaware of any statistics for outstanding amounts. (The banks themselves hopefully have an idea regarding their own exposure.) One could plausibly argue that these pre-approvals are a causal factor, but one runs into the lack of public measurement of the variable.

It is a mistake to look at the time series of mortgage debt and house prices and do disembodied statistical analysis on them. We know what the relationship is between the variables; it is just the aggregate of individual transactions. Therefore, a statistical analysis of "causality" is ignoring known dynamics. The relationship is instead somewhat mystical: current credit standards limit households' ability to bid up house prices, and this shows up in the growth of house prices and mortgage debt. Unfortunately, "credit standards" is a somewhat fuzzy concept. It can be quantified in the context of a single model of lending, but it is unclear whether a single model can capture observed behaviour.

I will return to the effect of credit standards in followup article(s) that will discuss the Canadian housing bubble. (As a spoiler, the fact that the hockey stick showed up when it did is not a surprise...)

Portfolio Allocations in Models

I will conclude with a discussion of how this fits into economic models. One of the standard complaints about mainstream economics by some heterodox economists is that "mainstream economics ignores the financial sector." Given what happened in 2008 (a global crisis that was the direct result of financial sector shenanigans), this seems self-evidently stupid. Although I am not a fan of "mainstream" economics, I am unsure whether we really need to model the financial sector itself. Any mathematical model of the economy will have to make some simplifying abstractions; and throwing out an explicit model of the financial sector is one of those useful abstractions. Instead, we could just implicitly model it via the effect on credit rationing.

If one looks at the stock-flow consistent models in Monetary Economics by Godley and Lavoie, the balance sheets of the financial sector are modelled. However, it is unclear how much value is added by doing so. For example, in the simplest model with multiple financial assets (model PC), the central bank is modelled as part of the allocation between interest-free money and Treasury bills. (Model PC is discussed in Sections 5.5 and 5.6 of An Introduction to SFC Models Using Python.) However, the detail associated with the central bank balance sheet is superfluous. We get exactly the same model outcomes if just specify an exogenous interest rate, and the private sector allocates between bills and money; the role of the central bank in the model is to accommodate that portfolio choice. We can make the models more complex by adding more assets, but at the end of the day, the financial sector has to accommodate the portfolio balance decisions of the other sectors.

When we look at the blob labelled "Financial Sector" in the circular flows figure (at the top of the article), the assumption is that the financial sector will act to accommodate the inflow/outflow associated with the mortgage borrowing. If the new flows would result in balance sheets that other entities are not happy with, the assumption is that prices have to shift. For example, imagine that the home seller wants to invest the proceeds in the stock market, instead of a fixed income fund (that would presumably be in a position to purchase the securitised mortgage). In order to patch up the flow mismatch, equity prices would presumably rise versus fixed income. This will then trigger some equity selling by some entity, which allocates the proceeds into fixed income (and thus being able to provide the funding for the home buyer). As our mainstream economists friends would say, it's all a question of the equilibrium prices shifting, and it all happens simultaneously within an aggregated model (so don't try explaining it with fables about flows).

Turning to the mortgage market, the role of the lender is to ration credit. In the real world, we obviously cannot rely on people to assess their own credit quality. Borrowing money that you have no real intention to repay is a time-honoured business model, and some practitioners have managed to do quite well for themselves. In a mathematical model, we cannot easily capture such behaviour, and so the credit capacity of an entity will be a function of some variables. In such a framework, we might as well assume that the borrower is the entity making the decision; the only role of the lender is as the implicit source of the function. (In an agent-based model, one could more easily attempt to model both sides of the transaction, but agent-based models create other modelling problems.)

This matters to how we interpret the hockey stick in Canadian mortgage borrowing. Did it happen because the Canadian financial sector woke up one day and decided to ramp up their mortgage volume, and the household sector had no choice but to borrow more? Not really. As I will discuss in the follow up article(s), what the financial sector did (or more accurately, what the Canadian Federal government did) was to loosen credit standards for mortgage insurance. This gave many households the new ability to lever up their balance sheets, and that is exactly what they proceeded to do. The role of the financial sector was to accommodate those households' choices.

This in turn creates a tension between thinking about models, and thinking about policy implications. Just because we want to model a housing bubble as the result of household choices (rather than the result of financial sector actions), there are almost no policy implications that result.

One could imagine someone arguing that since the build up of debt was the result of optimising choices by households, it would be sub-optimal policy to attempt to thwart those decisions. This could then result in the "neoliberal" laissez-faire attitude that summarises actual policy decisions over the past couple of decades. However, the hidden assumption is that households are making an intelligent decision that is in their best interest. Conversely, one could take the paternalistic view that we know that the housing bubble is what households want to do; we just don't let them, because if they do so, things will end very badly. For anyone who has not been indoctrinated by an economics training, such an attitude is not that controversial. Laws are generally passed to stop people from doing things that they want to do.

We see this in the goings on in Canada. Policymakers have been lecturing households for years regarding their rather lackadaisical attitude towards their debt levels. These lectures have been roughly as successful as my lectures to my cats regarding the timing of cat treat distribution. The fact that the household sector is the driver of borrowing makes the problem more intractable; it is a lot easier to sit down with a few bank CEO's and knock some common sense into them than it is to try to convince a large number of people to do anything that they do not want to do.

(c) Brian Romanchuk 2018