Mortgage Barometer depicts impact of the slow moving bad debt cycle

Posted On Thursday, 05 May 2016 21:25 Published by
Rate this item
(0 votes)

Perhaps one contributing factors to the formation of residential property “bubbles” or “overshoots” across the world over the years is the lengthy leads and lags between the start of a “negative” economic event, such as an interest rate hiking cycle, and the start of a rising trend in bad debt levels.


These lower turning points in the bad debt cycle are particularly dangerous periods for mortgage lenders. This is because, after interest rates begin to rise, it is often the case that for a lengthy period of time one doesn’t see any noticeable rise in bad debt/arrears levels.

During this period, the risk is that the mortgage lending sector can begin to mistakenly believe that it is “bullet proof”, or that the risks are lower than it previously thought. After all, a supposedly negative economic event has taken place but little or nothing bad has happened to mortgage loan arrears levels. The conclusion can possibly be that lending has been of such a high quality that home loans books can withstand these negative events. Recent times possibly are a case in point.

Economic growth has been weakening (although not in recession yet), and interest rates have been rising, but very little in terms of rising arrears/non-performing loan levels has been evident yet.  

There are a few possible reasons for this.

1. Lending quality has indeed improved dramatically, reducing overall Household Sector vulnerability It is old news that, following the pre-2008 housing bubble, mortgage lenders took a major pounding in terms of bad debt levels. The response of the industry was to tighten credit criteria significantly, focusing to a far greater extent on quality mortgage borrowers. This change shrunk the size of the national Household Sector Mortgage Debt level dramatically in real terms, from a peak of 49.2% of Disposable Income in the 1st quarter of 2008 to 34.83% by the 4th quarter of 2015.

This decline in turn contributed greatly to a significant decline in the overall Household Debt-to-Disposable Income ratio, lowering the vulnerability of households as a group to economic pressures, just in time given South Africa’s apparent entry into the long term “stagnation phase” of the economic super-cycle. The Household Sector Debt-to-Disposable Income Ratio was down to 77.8% by the end of 2015, from an all-time high of 88.8% at the beginning of 2008. 

This is a crucial improvement, but at 77.8% the ratio still remains on the high side, and we believe that further decline would be desirable.

So mortgage lending quality improvements since 2008 must surely play a key role in containing mortgage arrears levels in recent times, while interest rates have wound there way gradually higher. However, while this can contain levels of bad debts during the interest rate hiking cycle, we do not believe that it can alter the direction of the trend.

In other words, if interest rates and the all-important DebtService Ratio (which we shall discuss later) rise, mortgage loan arrears levels should ultimately rise, albeit by a smaller magnitude than would have been the case should household indebtedness have been significantly higher.

2. SARB interest rate hiking has been particularly slow this time around, giving households far more time to adapt than in previous cycles Important, too, perhaps, in “delaying” the onset of rising mortgage arrears, has been the “snails” pace of Reserve Bank (SARB) interest rate hiking. If one looks at the previous 3 interest rate hiking moves, 1998 brought about 7.25 percentage points’ worth of rate hikes in less than 2 months, 2002 brought 4 percentage points in about 9 months, and 2006-2008 saw 5 percentage points in approximately 2 years.

The current interest rate hiking cycle, by comparison, has seen a mere 2 percentage point’s worth of rate increases in over 2 years, having started back in January 2014. This slow pace of rate hiking must surely count for something, as households have had far more time to adapt their finances to rate hiking than was the case in previous interest rate hiking cycles.

3. “Natural” Lead and Lag times between the start of interest rate rises and the start of arrears increases. Finally, though, we need to ponder the lengthy leads and lags in the cycle of bad debt, and not allow ourselves to be fooled into thinking that we can escape totally unscathed this time around.  As already mentioned, interest rates began to rise slowly in January 2014.

This, in turn, began to drive the allimportant Household Sector Debt-Service Ratio higher. The Household Debt-Service Ratio is the Cost of Servicing the entire Household Sector Debt Burden (Interest only), expressed as a percentage of Household Sector Disposable Income. It is arguably one of the best single predictors of bad debt trends.

Its level depends on the combination of the level of Debt relative to Household Sector Disposable Income, as well as on the levels of interest rates at which households borrow.  The magnitude of the rise in the Debt-Service Ratio has indeed been curbed by a healthy decline in the Household Debt-to-Disposable Income Ratio, but the fact is that the Debt-Service Ratio has nevertheless risen in recent years.

We have a long data history for the Debt-Service Ratio. The only key indicator of credit health for which we have a corresponding long history is the Number of Insolvencies, published by StatsSA. We therefore use these 2 data series to examine the “natural” leads and lags at upper and lower turning points. There are considerable lag times between bottom as well as peak turning points in the Debt-Service Ratio and in the number of insolvencies. However, it appears that the lag at the bottom turning points in the cycle is considerably longer than at the peak turning points. 31

The grey areas on the chart illustrate the lag time between where the Debt-Service Ratio begins to rise “noticeably”, until where insolvencies start to rise “noticeably”, “noticeably obviously being subjective. Bottom turning points in the cycle, and the start of rising trends, are not always as clear as the upper turning points in the cycle. 

The 1st period highlighted was where the Debt-Service Ratio began to rise in the 2nd quarter of 1987. The lagged response before “take-off” by insolvencies was 8 quarters (2 years), beginning in 1989. The 2nd period highlighted saw the DebtService Ratio start to rise noticeably from the 1st quarter of 1995.

The noticeable response from insolvencies started around the 2nd quarter of 1997. During last decade’s housing and consumer boom, the big lift-off in the Debt Service Ratio really only started in the 4th quarter of 2005 (although it had been rising more gradually prior to that). Insolvencies appeared to started that steeper rising trend in the 2nd quarter of 2007.

So the general lag time at the lower turning point of the cycle appears to be not far from 2 years.  By comparison, the lag time from the peak turning point in the Debt-Service Ratio to the peak in the turning point in number of insolvencies appears to be a significantly shorter time. The 1st of the last 4 major Debt-Service Ratio peaks was in the 1st quarter of 1985. Insolvencies peaked 5 quarters later.

The Debt-Service Ratio peaked again in the 3rd quarter of 1991, and insolvencies peaked 4 quarters later. The next big Debt-Service Ratio peak was in the 4th quarter of 1998, and the lag time to insolvencies peak was a mere 2 quarters.

The last big Debt-Service Ratio peak was in the 3rd quarter of 2008, and insolvencies hit peak levels 3-4 quarters later The apparent lengthier “natural” lag time at the trough, compared to at the peaks, can therefore lull us into a false sense of security if we aren’t careful. Early in 2016, we passed the 2 year mark from when interest rates began to rise, and which caused the DebtService Ratio to begin to rise noticeably. That would suggest that we should begin to see some rise in the number of insolvencies in 2016 statistics, as they gradually emerge.

Time will tell. Admittedly, we probably shouldn’t expect, at this stage, to see anything more than a mildly rising trend, because the rise in the Household Debt-Service Ratio from 2014 to date has been moderate in magnitude and gradual.  Interest rates, at 10.5% Prime Rate, and a Debt-Service Ratio at 9.7% (up from 8.6% at the end of 2013 but still far below the 14.4% peak of 2008) are not painfully higher Nevertheless, some deterioration in Household Sector credit health is expected in 2016.

Are there any signs of looming credit health deterioration in the mortgage market yet? 

So now the question is are we seeing any signs of a turn towards a rising trend in household sector mortgage arrears yet? The answer is “possibly early signs”.

National Credit Regulator (NCR) data has hinted at a possible turn in the number of Household Sector mortgage accounts in arrears, when expressed as a percentage of total number of Household Sector mortgage accounts. After a broad multi-year declining trend, from 13% of total accounts in the 1st quarter of 2009 to 8.3% by the 4th quarter of 2014, the percentage of mortgage accounts in arrears had risen slightly to 8.7% of total accounts by the 4th quarter of 2015, reflecting a sideways to slightly upward move during last year.

When we examine the value of Household Sector mortgage arrears, as a percentage of the total value of Household Sector mortgage accounts, we see that the trend turning points appear to lag that of the number of accounts. The value of Household Sector mortgage arrears only commenced a noticeable declining trend in 2010, from 15.8% of total value of accounts in the 1st quarter of 2010 to 8.2% by the final quarter of 2015, still appearing to trend slightly downward as at the end of last year.

Why would arrears by number of accounts appear to turn the corner faster than by value? Possibly because the lower end of the residential market begins to see a pick up in financial pressure earlier than the higher priced end (and finds relief quicker after the interest rate cycle has peaked), therefore causing relatively more lower value mortgage loans to enter the arrears categories first. Is such a theory plausible? In an attempt to find support for it we turn to indicators of financial stress in the residential property market, as obtained via the FNB Estate Agent Survey.

Here, the estimated percentage of “sellers selling in order to downscale due to financial pressure” can be useful. While the estimate can be a little volatile from quarter to quarter, the most recent 2 quarters’ estimates of 14% and 13% respectively are off recent lows of 11%, and the smoothed trend line that we include in the graph has just started to rise mildly from 2015.

Breaking the estimate up into our 4 Income Area Segments, indeed we see the Lower Income Segment (average house price = R960,000) appears to have started to lead the way higher. We have smoothed these time series with a Hodrick-Prescott smoothing function, and the Lower Income Area smoothed estimate has moved to above 16% in the 1st quarter of 2016.

This represents a relative  turnaround from a stage of 2013 when Lower Income Areas had, by a small margin, the lowest estimated percentage of our 4 Income Area segments.

The Lower Income Segment is followed upward by the Middle Income Area Segment (average price = R1.34 million), while the Upper Income and High Net Worth segments show the least movement to date. It makes sense that Lower Income Areas should be most sensitive to economic cycle fluctuations and interest rate moves, as the Lower Income end is arguably more heavily credit dependent, and has less in the way of financial buffers to cushion the “blows”.

We therefore believe that the Home Owner Market has shown very early signs of rising financial stress, picked up in our FNB Estate Agent Survey. This may just be starting to feed through into residential mortgage loan arrears if one examines arrears on a “number of accounts” basis as opposed to the more well-monitored “value of arrears” measure, with the Lower Income end leading the way, although a few more datapoints are required before drawing hard and fast conclusions .  

Examining rental market data for the percentage of tenants in good standing with rental payments (supplied by TPN), we have not yet seen a meaningful deterioration. However, at 85%, the percentage of tenants in good standing is slightly off the highs of 86% reached in 2013 and 2014, after having experienced a broad rising trend through 2008 to early-2013 from a low of 71% back in the 2008/9 recession     


Examining our various mortgage and residential market data, therefore, there would appear to be very early signs of financial stress beginning to rise, which should ultimately translate into rising mortgage arrears levels. The FNB Estate Agent survey suggests that the lower end may have begun to lead the way to higher levels of financial stress-related selling of homes.

A lower end move may be the reason why the number of mortgage accounts in arrears looks like beginning to rise, as a ratio of total accounts, while the value of mortgage arrears hasn’t (i.e. possibly due to greater growth in the number of lower value arrears accounts. 2 years into the rising interest rate and rising Debt-Service Ratio cycle, the lag is about long enough for us to expect to begin seeing a mildly rising trend in financial stress-related home selling and mortgage arrears. Such a lengthy lag time may not be out of place, especially given that the pace of interest rate hiking has been “ultraslow” and moderate in magnitude to date.

In addition, mortgage lending has been significantly more conservative than in the pre-2008 boom years, resulting in higher quality of loans being put on the book, and a significant real decline in the value of mortgage books. It must be emphasized that all indicators point to still-low levels of financial stress in the residential market, despite hints at the start of a deterioration, and one should probably not expect this cycle to bring about an extreme deterioration given the small magnitude of interest rate hiking. Nevertheless, the mortgage lending sector can’t expect to escape an interest rate hiking cycle, in a long term stagnating economic growth environment, totally unscathed.  

In our forecasting we define “non-performing” loans as those in arrears for longer than 90 days, using NCR data. From what we believe to have been the low point in 2015, 2 years after the Debt-Service Ratio bottomed in 2013, we project the value of non-performing mortgage loans to rise moderately through our forecast period from 2016 to 2018, from 3.4% of the value of total mortgage loans in 2015 to 4.76% by 2018. This is based on a forecast of mild further interest rate hiking, to where Prime Rate peaks at 11% later this year, taking the Debt-Service Ratio to just above 10% 

Last modified on Monday, 09 May 2016 03:58

Most Popular

Deeds Office reopening good news for sellers and buyers of property, as well as the property industry

May 01, 2020
Andrew Golding Golding Property Group
Deeds Office reopening good news for sellers and buyers of property, as well as the…

Wide-spread implications for South Africa’s real estate market following COVID-19

May 05, 2020
JLL, one of the world’s leading real estate investment and advisory firms, today released…

Deeds office reopen their doors to the public

May 09, 2020
Carlize Knoesen
The Department Agriculture, Land Reform and Rural Development has announced the reopening…

Relooking green buildings in the future of the covid-19 pandemic

May 01, 2020
Brett Chrystal
The Covid-19 lockdown has offered us an unexpected opportunity to reflect on our…

SA REIT appoints Joanne Solomon as its first CEO

May 05, 2020
Joanne Solomon new CEO SA REIT Association
With her wealth of financial and property sector experience, Joanne Solomon has been…

Please publish modules in offcanvas position.