Gearing Ratios: Trading Company vs Bank

[Grunter]

I see your point Grunter, but I do not agree with your conclusion. IMO a higher gearing ratio always means more risk for any specific company.

Here is a paper that you may be interested in that demonstrates that there is no relationship between the gearing ratio and the risk of default in a bank.

http://www.szgerzensee.ch/fileadmin/...rs/wp-0204.pdf

[Snoopy]

Here is a paper that you may be interested in that demonstrates that there is no relationship between the gearing ratio and the risk of default in a bank.

http://www.szgerzensee.ch/fileadmin/...rs/wp-0204.pdf

Good on you for injecting some intellectual rigour into the debate Grunter.

The introduction to this paper contains the following sentence:

"It is largely undisputed that, everything else being constant, a bank's probability of default decreases with the level of capital - a simple buffer stock effect."

This is the point I was making, and the paper does not attempt to dispute it.

However, the paper goes on to talk about the case where: "everything else is not constant."

The main variation as I read it, talks about bank shareholder pressure to create competitive returns. While more capital supporting the same level of business will generate a lower ROE for the bank (everything else being equal) , a well capitalised bank may be incentivised to take on 'extra risky high profit deals' in an attempt to counteract the drag on profitability caused by having more equity on the books. Thus the net positive 'safety buffer'' effect of having more equity on the books, might be undone by the shady eighteenth floor ivory tower traders that the bank employs.

Critical in refuting or not this effect is how to measure 'risk'. The paper covers what happened to Swiss banks between 1990 and 2002. During that period there is no mention of any Swiss bank going bust. So the risk of going bust has to be measured indirectly by an indicator. Section 3 of the paper talks about how to measure risk. Risk is talked about as the 'volatility of bank assets' by valuation. There is an acknowledgment that his is difficult to measure directly, and that the study used 'a call option on the value of the assets of the bank' to measure it. Any option is simply a money weighted guess by market players, themselves most likely bank employees. This injects yet another layer of speculation over the true underlying data that you are trying to measure.

Ths problem with this study, as I see it, is that the risk being measured is not the risk I am concerned with. If, as a banker, my underlying loan portfolio suddenly had an increase in value of 20% of the underlying assets, then this would be a very good thing. Suddenly my loan would be backed by a much more valuable hard asset. Yet under the 'risk' is 'volatility' prescription, such an sudden increase in assets would be very bad because it would increase asset valuation volatility!

I put it to you Grunter that the chosen measure of risk talked about in the paper doesn't reflect actual 'shareholder risk' or 'bank default' risk. The 'employee behavioural angle' of risk taking behaviour being incentivised is interesting. But the conclusion that volatility in the asset portfolio is a useful indicator of the likelihood of a bank loan defaulting over its term is IMO flawed. Consequently the conclusions drawn by the authors of the paper must be open to question.

SNOOPY

[Grunter]

Good on you for injecting some intellectual rigour inot the debate Grunter.

The introduction to this paper contains the following sentence:

"It is largely undisputed that, everything else being constant, a bank's probability of default decreases with the level of capital - a simple buffer stock effect."

You will note the paper reaches this conclusion:

"we do not find a significant relationship between the default probability and the capital ratio"

You argue that the method of quantifying risk is flawed and point to volatility not being an appropriate measure of risk? Volatility is the very definition of risk. If the assets appreciate, yes, they have increased in value suddenly, but that does not mean volatility has increased. Volatility comes about when values jump around (up AND down) about a mean. If prices are only going one way (up), the mean is moving upwards as well and the time series of asset values are said to be trending. It's just like a stock price - volatility only increases when the price is moving around more, rather than in one particular direction.

The paper argues that the best approximate for the volatility of the bank's assets is the volatility of the bank's returns and hence reflected in the stock price. Unless you work in the bank and calculate the probability of default on each and every loan in the bank's portfolio, I don't think you are going to get a better indicator. In fact, you may see that the distribution of default risk may in fact be normal, so that you can make an overall assumption of default risk from the bank's returns anyway.

I think you get yourself into trouble when you attempt to granulise your analysis by considering a single loan situation, rather than looking at the overall loan book.

[Snoopy]

Snoopy, I think the first thing you need to consider is what "risk" is. From the firm's point of view, the risk you are considering is the risk of not being able to meet their debt obligations, or default risk.

Grunter, this is the first comment on risk that you made on this thread, I comment that I agree with.

You argue that the method of quantifying risk is flawed and point to volatility not being an appropriate measure of risk? Volatility is the very definition of risk. If the assets appreciate, yes, they have increased in value suddenly, but that does not mean volatility has increased. Volatility comes about when values jump around (up AND down) about a mean. If prices are only going one way (up), the mean is moving upwards as well and the time series of asset values are said to be trending. It's just like a stock price - volatility only increases when the price is moving around more, rather than in one particular direction.

I think you get yourself into trouble when you attempt to granulise your analysis by considering a single loan situation, rather than looking at the overall loan book.

I contend that there are many possible definitions of risk. Volatility is but one definition of risk. "Volatility= Risk" is favourable in a mathematical sense becasue it is (relatively) easy to deal with if you are working within the mathematics of risk. But I want to go back to your first statement Grunter, talking about 'default risk'.

Now let's take an example. Rather than a 'single example', for which you have criticised me examining in the past, let's talk about a wider NZ based risk that has been in the bankers sights of late: dairy farm risk.

Dairy farm risk is not the same for every dairy farmer. The consensus though, is that with typical levels of borrowing and typical production yield, the NZD payout price per kg of milk solids needs to be about $NZ5.50 for our 'average farmer' to break even, and meet 'debt obligations'. We can discuss this dairy industry case with some confidence, because unlike the Swiss bank study, we have a direct and widely published indicator of output receipts: the monthly Fonterra Milk Auction Results. There is also a milk futures market that is an industry and speculator view of where milk solid prices might be headed in the future.

Based on these auction results Fonterra issues its 'best guess' as to what the milk payout will be for the season. The price received is largely out of farmers hands. Farmers can control their input costs though. And you can measure how good a farmer is, in a financial sense, by measuring how well they can control their input costs, and therefore calculate a farmers margin by taking the projected annual farm payout and subtracting from that those input costs. Input costs include a farmers debt obligations.

My contention is that the best measure of 'dairy farm risk' (the risk of a dairy farm not being able to meet theri debt obligations) is to measure dairy farm input costs.

Your contention is that "risk = volatility". So can you please explain to me how 'volatility' is used to measure dairy industry default risk?

SNOOPY

[Snoopy]

You argue that the method of quantifying risk is flawed and point to volatility not being an appropriate measure of risk? Volatility is the very definition of risk. If the assets appreciate, yes, they have increased in value suddenly, but that does not mean volatility has increased. Volatility comes about when values jump around (up AND down) about a mean. If prices are only going one way (up), the mean is moving upwards as well and the time series of asset values are said to be trending. It's just like a stock price - volatility only increases when the price is moving around more, rather than in one particular direction.

I contend that there are many possible definitions of risk. Volatility is but one definition of risk.

Your contention is that "risk = volatility". So can you please explain to me how 'volatility' is used to measure dairy industry default risk?

I hope Grunter is just on holiday and he (or anyone!) will soon return with the answer to my question. Perhaps not though, because to answer my question I think will be difficult.

The situation as I see it is as follows.

A banker lending to the dairy industry today will want to see dairy revenue become greater than dairy input costs. Of most concern is the milk price trend, or more exactly the future milk price that can be extrapolated from that trend. If the input costs exceeds the revenue received, then long term the industry is not sustainable. Farming commodities have a long history of price 'ebb and flow' though. So the smart banker will not immediately foreclose on a farm just because input costs exceed revenue. Instead, such farms will go on a 'watch list'. If the milk price trend is down, and revenue received is below the cost of sales, then this is the worst situation. The banker will be looking for the commodity price to bottom. Once a commodity price has stabalised, then our banker will be able to assess how long the farm has got until its banking covenants are broken. Where there is some certainty, a banker will feel comfortable lending. Not having certainty is what bankers don't like!

'Volatility in milk price' will be the most unsettling thing for a banker. If the milk price crashes, then suddenly recovers before going down, only to recover again, then our banker will not be able to make any decision. What our banker needs to know is:

1/ where the milk price is today AND
2/ whether the milk is trending up or down. AND
3/ how the current price matches the breakeven cost for the dairy industry.

If our banker knows the three things above, then he can assess industry risk.

However, if the price is bouncing around seemingly with no direction, typically what you see with volatility then our banker can't detect a trend, can't detect a likely bottoming or peaking of price and therefore can't assess how long a dairy business can use its equity to ride out the bad times. Volatility is largely useless to our banker in assessing default risk at any level. Volatility is not measuring any of the factors that determine the future viability of the dairy industry. Thus when the previously mentioned academic study suggests that 'high volatility', what Grunter refers to as jumping up AND down, is not correlated with 'default risk' this is exactly what I would expect. If your chosen indicator (volatility) is measuring something completely unconnected to 'default risk', of course there will be no correlation observed!

This doesn't mean you can't measure default risk though. It just means that IMO volatility is completely the wrong tool with which to do the measuring!

SNOOPY

[Grunter]

Snoopy, again, respectfully I don't think you properly understand the definition of risk.

I've included another paper that talks about how default risk is measured. From it, you should see how volatility and the risk of default are inextricably linked.

http://www.econ2.jhu.edu/People/Duffee/rfs.pdf

[Snoopy]

Snoopy, again, respectfully I don't think you properly understand the definition of risk.

I've included another paper that talks about how default risk is measured. From it, you should see how volatility and the risk of default are inextricably linked.

http://www.econ2.jhu.edu/People/Duffee/rfs.pdf

Grunter, if I read the paper correctly, the authors are studying corporate bonds, the vast majority investment grade. In particular the study looks at the market interest rate of those bonds as it varies in time with a one month sampling frequency for the data. Attempts are then made to see if this data can be used to correlate with the default risk of the assiciated individual company for each bond considered.

A default is a default after it has happened. But there is a long standing practical problem of determining the mathematical boundaries for when a default happens - the covenant that triggers the default can vary between companies in practice. This paper attempts to get around this problem by defining default as a "two-factor square-root diffusion model."

A "two-factor square-root diffusion model", from what I can figure, is in general terms, modelling (equations 3a and 3b) a kind of 'decay behaviour' (By this I mean you start with a constant value and subtract a time series function around variable 's'). The 'initial state' dissipates in proportion to the function of 's' entities decaying, plus a variable term related to the square root of 's' multipled by the standard deviation from some central condition. In mathematical terms the form is:

Change = [Constant - Decay Function] + [Standard Deviation] x [Square Root of Decay Function] x [Another Constant]

This equation form is entirely selected by the authors, and already includes a 'variability term'. Thus far from 'proving' that variability and its triigger of volatility is related to company default, the paper appears to set up this prospect by 'definition' at the outset.

Then, in a 'huge surprise' ;-), this 'defintion of default' is best actuated by another function, the bouncing up and down of bond interest rates which introduces a 'volatility' that satisfies the variability trigger of the model! I may have interpreted all this incorrectly. But it looks to me Grunter, like a circular argument you are running. Define a 'default' as a 'function of variabilty', then introduce a 'volatility' that triggers the 'variabilit'y that 'proves' your case.

I am struck by the following comment in the introduction to the reference paper

"Much of the literature follows Merton (1974) by explicitly linking the risk of a firm’s de-fault to the variability in the firm’s asset value. Although this line of research has proven very useful in addressing the qualitatively important aspects of pricing credit risks, it has been less successful in practical applications."

Have you considered that the reason many of these studies have 'failed to work in practice' (for individual companies) is that the underlying premise of the model being studied is wrong?

Are you able to articulate, without referencing an academic paper, exactly why you feel 'volatility' is related to the 'risk of bankruptcy' for any particular investment?

SNOOPY

[Grunter]

Grunter, if I read the paper correctly, the authors are studying corporate bonds, the vast majority investment grade. In particular the study looks at the market interest rate of those bonds as it varies in time with a one month sampling frequency for the data. Attempts are then made to see if this data can be used to correlate with the default risk of the assiciated individual company for each bond considered.

A default is a default after it has happened. But there is a long standing practical problem of determining the mathematical boundaries for when a default happens - the covenant that triggers the default can vary between companies in practice. This paper attempts to get around this problem by defining default as a "two-factor square-root diffusion model."

A "two-factor square-root diffusion model", from what I can figure, is in general terms, modelling (equations 3a and 3b) a kind of 'decay behaviour' (By this I mean you start with a constant value and subtract a time series function around variable 's'). The 'initial state' dissipates in proportion to the function of 's' entities decaying, plus a variable term related to the square root of 's' multipled by the standard deviation from some central condition. In mathematical terms the form is:

Change = [Constant - Decay Function] + [Standard Deviation] x [Square Root of Decay Function] x [Another Constant]

This equation form is entirely selected by the authors, and already includes a 'variability term'. Thus far from 'proving' that variability and its triigger of volatility is related to company default, the paper appears to set up this prospect by 'definition' at the outset.

Then, in a 'huge surprise' ;-), this 'defintion of default' is best actuated by another function, the bouncing up and down of bond interest rates which introduces a 'volatility' that satisfies the variability trigger of the model! I may have interpreted all this incorrectly. But it looks to me Grunter, like a circular argument you are running. Define a 'default' as a 'function of variabilty', then introduce a 'volatility' that triggers the 'variabilit'y that 'proves' your case.

I am struck by the following comment in the introduction to the reference paper

"Much of the literature follows Merton (1974) by explicitly linking the risk of a firm’s de-fault to the variability in the firm’s asset value. Although this line of research has proven very useful in addressing the qualitatively important aspects of pricing credit risks, it has been less successful in practical applications."

Have you considered that the reason many of these studies have 'failed to work in practice' (for individual companies) is that the underlying premise of the model being studied is wrong?

Are you able to articulate, without referencing an academic paper, exactly why you feel 'volatility' is related to the 'risk of bankruptcy' for any particular investment?

SNOOPY

Snoopy,

Most banks are listed, and also have debt issuances. In sophisticated markets, most also have Credit Default Swaps on these bonds. This is the best way to price the risk of default in the bank.

You seem to think that Banks are an entirely different beast - they are not.

You also seem to not know what parts of the paper to read - you don't read what the paper is proposing - just read the theory parts. I'm not trying to propose ways of measuring risk - I post the papers so they explain the theoretical underpinnings of the argument to make clearer to you.

Essentially volatility is related to the risk of bankruptcy as the risk of bankruptcy is defined as when the value of the bank's assets fall to a defined amount (zero/below solvency requirements whatever). As the value of the assets fluctuates through time, this means that the value of the assets is volatile. Therefore it is easy to see that low volatility means that it is less likely the bank's assets will drop below the bankruptcy benchmark, and more volatility means that it is more likely the benchmark will be reached.

I fail to see how you haven't grasped this concept?

[Snoopy]

You also seem to not know what parts of the paper to read - you don't read what the paper is proposing - just read the theory parts. I'm not trying to propose ways of measuring risk - I post the papers so they explain the theoretical underpinnings of the argument to make clearer to you.

You may be right here Grunter. Those papers you referenced are very 'information dense'. If you wanted to point be to a particular part of the paper then you should have felt free to do so. I will have a look at it again. But I reckon that if I was looking for the 'theoretical underpinnings', then just reading the theory parts should have been a good place to start!

Essentially volatility is related to the risk of bankruptcy as the risk of bankruptcy is defined as when the value of the bank's assets fall to a defined amount (zero/below solvency requirements whatever). As the value of the assets fluctuates through time, this means that the value of the assets is volatile. Therefore it is easy to see that low volatility means that it is less likely the bank's assets will drop below the bankruptcy benchmark, and more volatility means that it is more likely the benchmark will be reached.

I fail to see how you haven't grasped this concept?

Thanks for the more explicit explanation. But the one point you haven't answered is the one you put so well yourself previously.

You argue that the method of quantifying risk is flawed and point to volatility not being an appropriate measure of risk? Volatility is the very definition of risk. If the assets appreciate, yes, they have increased in value suddenly, but that does not mean volatility has increased. Volatility comes about when values jump around (up AND down) about a mean. If prices are only going one way (up), the mean is moving upwards as well and the time series of asset values are said to be trending. It's just like a stock price - volatility only increases when the price is moving around more, rather than in one particular direction.

I totally get the bit about when a bank's assets fall below a defined amount, there is a problem. But as you so eloquently hinted at above, this is a problem if it happens at the end of a 'trend' (a downward trend being the potential problem). If a bank's assets go below a certain value due to 'volatility' (could be a problem), then you can expect those assets to bounce back up in value again due to that same 'volatility' (actually not a problem).

So while I agree with you on the issue at stake:

"bank's assets fall to a defined amount"

I don't agree that the threat here is 'volatility'. And quoting from your previous comments above, from 21st September 2016, it seems you agree with me!

SNOOPY

[Snoopy]

Let's say our bank is making a loan to a company producing an automated production line for making fridges in China. Suddenly NZ gets foot and mouth disease, consumer confidence in NZ collapses, and NZ consumers as group need their deposit money to pay down debt of other NZ members of this consumer group. Now I think we can agree there is very likely no connection between building a production line for making washing machines (or fridges ;-) )in China and foot and mouth disease on New Zealand farms. But because these two have nothing in common, this introduces a 'funding risk' for the bank loans. A risk where one side wants out of the loan contract for reasons completely unrelated to the project the funds are being used for.

'Liquidity', the mismatch between the needs of depositors and borrowers (such as in the example above) , is a side issue in the overall context of this thread. Yet I feel it is worthy of some comment.

In a contractual sense, banks tend to have enormous liquidity problems. In a practical sense they tend to have no problems. This is because depositors tend to:

1/ Roll over their debenture investments AND/OR
2/ Not suddenly pull out all their cash funds that are on call.

Furthermore the banks can influence depositor behaviour by offering higher interest rates that correspond to periods where the bank wants to retain more funds.

There is further 'fund relief' available too. A bank can simply raise more equity through something as simple as a dividend reinvestment plan. So despite the dire contractual picture on most bank books, in reality getting more funds to pay out depositors is not an issue. A extension of that argument is that studying bank liquidity is a waste of time.

No bank CEO knowingly makes a loan that he/she knows will be difficult to repay in a timely manner. Yet we do know that often the first step in a bank loan going bad is a problem with liquidity. Liquidity issues are often resolved by bringing in new capital from somewhere else. This is what I call a 'sideways solution', because bringing in new capital does not necessarily solve the underlying liquidity problem. Rather, the new capital usurps the original problem so the original underlying liquidity issue becomes redundant.

Here then is the enigma for investors concerned with bank liquidity. If the most common way to fix an emergency 'liquidity issue' has nothing to do with manipulating the timing of cashflows of depositors and lenders, does it make sense to study liquidity at all?

SNOOPY