Comments on Bank of England’s Quarterly Bulletin on Household Debt and Spending

The Bank of England report “Household debt and spending” stated that “it is difficult to evaluate whether debt has had any impact on UK household spending using aggregate data alone. Indeed, UK consumption grew at roughly the same rate between 1999 and 2007, when debt was rising rapidly, as it did between 1992 and 1998, when debt did not increase relative to income. This, together with the fact that increases in household debt were largely matched by a build-up in assets, is consistent with the suggestion that increases in debt did not provide significant support to consumption.”

First of all household expenditure did not grow at roughly the same rates over this period:


I have used the start point for the analysis as the peak of the previous economic cycle given that part of the growth in the early 1990s would have been due the rebound in consumption from this earlier recession.  In fact, we can see that growth initially accelerated to Q4 1994, but then set off again on a second substantial leg that peaked between Q1 2000 and Q4 2001.

I wanted to explore this further here:

The following chart shows the annual change in unsecured credit, a debt dynamic excluded from the B of E analysis, as a % of annual change in GDP (on a rolling basis).   The large swings during the crisis are excluded for the sake of this analysis. 


Unsecured debt was rising at fairly heady clip during the 1990s.  In fact, the unsecured credit commenced its upward trajectory (based on available data) in the 3rd quarter of 1993 well below the left hand shoulder of the blue data line in the first chart.

What was not rising at such a heady clip was secured lending, that is until late 1995/early 1996 when the uptrend started:


But there are good reasons for this.  The UK had already succumbed to a housing boom and subsequent collapse in the late 80s and early 1990s.  But, importantly, for consumer expenditure and economic growth, interest rates fell significantly:


Much of the growth in household expenditure in the early part of the 1990s was a recovery/bounce back from the early 1990s recession, something which you cannot ascribe to consumer expenditure during the 1999/2007 period.  That said, a good part of the growth in expenditure in the latter half of the 1990s occurred during a period of rising secured and unsecured lending.


The following chart shows the cumulative increase in household expenditure less the cumulative increase in GDP from 1948 to 2013:


The late 1980s boom pushed household expenditure up relative to GDP growth and we can see a distinctive and substantial secondary leg up from 1994/1995 onwards.

The following shows the annual change of total sterling net lending to individuals:


And the quarterly change in unsecured net lending to individuals….which has bounced back of late:


And note that unsecured lending growth slowed from 2005 until the crisis…it looks as if rising house prices fuelled by secured lending growth facilitated a significant boom in home equity withdrawals during that period:


Interestingly while the ascent in debt to GDP has declined as the economy has recovered, debt to disposable income has plateaued reflecting the weak income growth scenario in the economy.


Post 2001 real household disposable income per head has been declining meaning that household expenditure growth had to increasingly dig into savings (or de facto borrowing), at least until 2008:



The statement by the B of E that it is difficult to discern the influence of debt via aggregate data ignores some important points: the impact of recovery and declining interest rates which would have boosted expenditure during the first portion of the period highlighted by the report, significant increases in unsecured lending during the first part of the period and the very large home equity withdrawals that occurred between 2000 and 2007. 

Household expenditure as a % of GDP remains elevated as does consumer debt relative to disposable income, given weak income growth, and relative to GDP. image

See also :

And some further comments on that report:


The report into “Household debt and spending” used the Office of National Statistics Living Costs and Food survey to assess a rather narrowly defined relationship between the sensitivity of consumption to debt levels. An assessment of the sensitivity of consumption to debt levels is useful but too narrow a frame to address the issue of debt and excess asset focussed money supply growth. The study also did not include unsecured debt in its analysis.

In the second half of the 1990s, households with mortgage debt to income ratios greater than 2 appear to have increased the share of their income spent on non-housing consumption by more than mortgagors with lower debt to income ratios.

Regression analysis confirms that households with higher debt levels made larger adjustments in spending after 2007, even after controlling for other factors. An econometric model in which households’ consumption is determined — in part — by their mortgage debt to income ratio,(2) can be used to estimate the impact of cuts in spending associated with debt on aggregate consumption since 2007. This estimate is constructed by taking the model’s prediction, for each household, for spending in a given year, and then subtracting what the model predicts they would have spent if debt had

Only focussing on “differences in spending patterns across households with different levels of debt” ignores the impact of increases in debt over time on economic stability. The line between debt, income, expenditure and economic growth is also a naturally blurry one: If I borrow and spend a loan on goods and services, this loan ends up as someone else’s income (whether it be revenue, wages, profits, dividends, taxation etc) so the lines between income, GDP and debt become blurred.

we are not able to observe the same households over time — only ones with similar characteristics — because the LCF Survey covers different households from year to year. This means that we cannot observe what the pre-crisis debt of individuals surveyed after the crisis was; and equally, we cannot observe how the debt of individuals surveyed before the crisis has evolved since then. Second, the measure of income used is net of mortgage interest payments, which means that reductions in mortgage rates after 2009 that lowered interest payments will have helped to cushion the squeeze in incomes for mortgagors.(1)

The fact that those with higher debts spent a higher amount of income on non housing consumption is also difficult to interpret. But we do know that household expenditure as a % of both GDP and personal disposable income was increasing during this time period which makes consumption and GDP even more exposed to interest rate and asset valuation/debt deflation risks.

But in my opinion, a large part of the debt problem itself was its increasing asset focus over the period.

The microdata analysis also implicitly assumes that most aspects of the economy were not affected by developments in household debt. Growth in debt could have had macroeconomic effects that may have fed back into consumption, for example, through its effects on employment, the public finances and asset prices. And, as explained in the first section, for some households to hold debt, others have to hold assets, and that could affect their behaviour. But attempting to evaluate either of these effects is beyond the scope of this article

The report’s analysis also ignored the impact of aggressive monetary policy during the crisis in terms of supporting asset prices which themselves supported consumer debt. The impact of debt defaults on consumption would likely have been much higher without QE meaning that much of the impact may still be latent.

Increases in debt levels are not in themselves bad things, but the period itself is one marked by falling interest rates, rising asset values and rising inequality as well as greater sensitivity of asset values to growth, inflation and interest rates.

If interest rates are falling because of weakening growth dynamics, then rising asset values, in response to falling IRs and increasing asset focussed money supply growth, is at odds with the long term economic fundamentals. Falling IRs, rising asset prices and growth supported by debt dynamics may well have supported GDP growth above levels the economy was capable of sustaining to justify balance sheet asset and debt values.

Leave a Reply