Using Credit Card Payments Data For The Public Good

Interesting post from the UK’s Office for National Statistics blog, which highlights the power of data analytics using anonymised  credit card payments data.

The intelligent use of data gathered by our leading financial institutions can result in faster, more detailed economic statistics.  Tom Smith describes how a joint event staged by ONS and Barclaycard illustrates the vast statistical potential of  anonymised  payments data.

“My job at the Data Science Campus brings many fascinating days as we work with organisations across government and the UK to unlock the power of data. One recent event particularly stands out.

Our experts from across ONS joined forces with analysts from one of the world’s biggest financial organisations to explore how commercial payments data could help tackle some of the UK’s biggest economic questions.

Following a successful knowledge sharing day at the ONS Data Science Campus, Barclaycard, which sees nearly half of the nation’s debit and credit card transactions, hosted a ‘hackathon’ at the state-of-the-art fintech innovation centre Rise. This brought together 50 economists, developers, data scientists and analysts to address three challenges:

  • How could payments data improve our understanding of regional economies?
  • Where could financial inclusion policies best be targeted?
  • How could we use payments data to create superfast economic indicators?

Over two days, the ONS and Barclaycard teams worked collaboratively – in some cases right through the night – to identify how the payments data could be used to improve our understanding of the economy. The traditional hackathon finish saw the teams ‘pitching’ their work to a panel of judges from across ONS and Barclaycard.

The winning team focused on building predictors and indicators that provide fine-detail information for trending economic changes. Even at this early stage of development, their work shows how bringing together card spending data and economic data held by ONS could improve the information available for policy & strategy decision makers to make timely economic decisions.

There is much work to be done to turn this demonstration into a working model. But one of the things that stood-out for the judges was the winning team’s roadmap for how to get there, including the development and data architecture needed for a successful prototype.

“We’re really excited to play a key role in helping to support a better understanding of UK economic trends and growth. The hackathon was a great event to harness the excitement and expertise created through our partnership with the ONS, and the winning teams have shown tangible evidence that payments data can indeed be used for public good.” – Jon Hussey, MD Data & Strategic Analytics, Barclaycard International

For the Data Science Campus, collaborations are all about knowledge exchange. They are an opportunity for us to access expertise in tools, technologies and approaches to data science from outside government, evaluate them in a safe environment, and share our learning across ONS and wider government.

It was inspiring to see the level of energy, drive and collaboration, and to pool ONS and Barclaycard skills into understanding how payments data can be used for public good. (And it is worth pointing out that no money changed hands and no personal data were involved. ONS is only interested in producing aggregate statistics and analysis.)

Our work with Barclaycard illustrates perfectly how the rich data held by partners outside government can improve our understanding of the UK’s economy. This is a key part of ONS’ Better Statistics, Better Decisions strategy, enabling ONS to deliver high quality statistics, develop and implement innovative methods, and build data science capability by tapping in to best practices wherever they may be.

New Home Sales Slide

According to the HIA, New Homes Sales Report – a survey of Australia’s largest home builders – there has been a fall in the number of new homes sold in 2017. New home sales were 6 per cent lower in the year to November 2017 than in the same period last year. Building approvals are also down over this time frame by 2.1 per cent for the year.

The HIA expects that the market will continue to cool as subdued wage pressures, lower economic growth and constraints on investors result in the new building activity transitioning back to more sustainable levels by the end of 2018.

This is a smaller down-turn than we anticipated and bodes well in terms of the likelihood of a modest and orderly reduction in new house building.

The story is not consistent across all of the states with Western Australia and Victoria providing the book ends on five very different stories.

In the middle of the year it looked like Western Australia had turned the corner after a significant decline in activity over three years, but the new financial year brought even lower results as more restrictive first home buyer policies were implemented.

At the other end of the market in Victoria, the expected slowdown in building activity has not yet materialised. Sales of new houses increased by 6.3% for the 12 months to November 2017 and approvals rose by a further 8.7 per cent in the three months to November compared with the same period in 2016.

Why the RBA is unlikely to cut interest rates

From Business Insider.

Australia’s housing market is cooling after years of rollicking price growth.

Annual price growth has halved since May, auction clearance rates sit at multi-year lows in Sydney and Melbourne and investor housing credit is declining, coinciding with tougher restrictions on interest-only lending from APRA, Australia’s banking regulator, introduced in March.

The slowdown in the housing market, coming on top of weakness in Australia’s household sector seen in Australia’s recent GDP report, has got many people questioning whether the Reserve Bank of Australia (RBA) should hike interest rates given the current set of circumstances, especially with inflationary pressures close to non-existent.

Rather than hiking interest rates, some have even floated the idea that the RBA may consider cutting interest rates given the sharp deceleration in the housing market.

George Tharenou and Carlos Cacho, Economists at UBS, played devils advocate on that front earlier this month, pointing to the chart below to show that when house prices weakened by a similar amount in the past, it has almost always resulted in the RBA cutting official interest rates.

The pair note that national price growth on a six-month annualised basis is currently running at just 0.7%, an important consideration given that over the past 30 years “when house prices over a 6-month period weakened towards flat or negative, the RBA cut within a few months in 7 of 9 cycles”.

While that’s not UBS’ official call, forecasting instead that the RBA will hike rates in late 2018 with the risks slanted towards a later move, it does pose the question as to whether the current weakness in the housing market will see history repeat.

To ANZ Bank’s Australian economics team, led by David Plank, the answer to that question is almost certainly no.

 “There has been quite a lot of focus on the current downturn in house price inflation, with some commentators pointing out that similar downturns in the past have been followed by RBA rate cuts,” the bank says.

“While this might be true, it ignores the key differences between this cycle and previous downturns.

“In particular, previous downturns in house prices followed a succession of RBA rate increases, which pushed mortgage rates sharply higher. Given that RBA tightening cycles typically impact a lot more across the economy than just house prices, we think it is difficult to argue that the slowdown in house price inflation was the primary reason for eventual rate cuts.

“We think a rising unemployment rate was far more important,” it says.

One look at the charts below adds credence to that view.

The first looks at the relationship between annual house price growth and mortgage rates. The latter, shown in orange, has been inverted and advanced by six months.

As opposed to what has been seen previously when house prices tended to decline following a series of interest rate hikes, in recent times, price growth has slowed despite mortgage rates remaining near the lowest levels on record, coinciding with tighter macroprudential restrictions on investor and interest-only lending from APRA.

“The current downturn in house prices has not come after a tightening cycle. Instead we think the most likely cause was the tightening in credit, though with a lag and interrupted by the impact of RBA rate cuts in 2016,” ANZ says.

In comparison, this next chart shows the relationship between the annual change in Australia’s unemployment rate to movements in the cash rate.

While not perfect by any stretch, when unemployment starts to lift, the RBA tends to cut the cash rate, and vice versus.

Australia’s unemployment rate has recently fallen to 5.4%, leaving it at the lowest level in close to five years, going someway to explaining why ANZ is forecasting that the RBA will lift the cash rate to 2% by the end of next year despite the slowdown in the housing market.

“In our view, a [housing] cycle driven by credit is likely to play out very differently from one driven by higher interest rates,” it says.

“Expecting the current housing cycle to play out like those caused by movements in interest rates, strikes us as likely to end in disappointment.”

Indeed, outside of the recent price deceleration caused by credit rather than mortgage rates, ANZ points to a variety of other housing market indicators that suggest there’s little need for the RBA to cut rates.

“The most recent data on auction clearance rates suggest some stability after a period of decline. If this broadly continues then we would expect house annual price inflation to stabilise in the low-to-mid single digits in 2018,” it says.

“Our forecasts have nationwide house price inflation slowing to zero in 2018, but this also includes the impact of the two RBA rate hikes we expect in 2018. If these don’t take place then we would expect less of a slowdown in housing inflation, probably to the low-to-mid single digits mentioned above.”

ANZ says recent strength in Australian building approvals data, supporting the view that credit cycles play out differently from rate hike cycles, provides further evidence why RBA rate cuts are not required on this occasion.

“In late 2016, when approvals were falling sharply, there were a number of dire predictions about what that would mean for housing construction and employment. But it has been clear for some time that the downturn in building approvals was shallower than in previous cycles,” it says.

“We think this is because this cycle was not triggered by higher interest rates. Instead, we think a more likely cause was the tightening in credit that began in 2015.”

According to the ABS, Australian building approvals rose by 0.9% to 19,074 in seasonally adjusted terms in October, leaving the increase on a year earlier at 18.4%. Private sector approvals for houses and other dwellings stood at 10,063 and 8,683, up 6.2% and 37.6% respectively from 12 months earlier.

Given the absence of weakness in other areas of the housing market, differing it from periods in the past when interest rates were cut, it helps explain why ANZ and the vast majority of forecasters believe that the next move in the cash rate will be higher, albeit not for many months.

Business Finance Still Skewed Towards Property

The final piece of the October 2017 lending finance data came from the ABS today. It is not pretty.  As usual we will focus on the trend series which irons out some of the statistical bumps.

Owner occupied housing lending excluding alterations and additions fell 0.1% in trend terms. Personal finance commitments rose 1.3%. Fixed lending commitments rose 2.2%, while revolving credit commitments fell 0.1%.

Total commercial finance commitments fell 1.1%. Fixed lending commitments fell 2.5% (which includes mortgage lending for investment purposes), while revolving credit commitments rose 3.7%.

The trend series for the value of total lease finance commitments fell 0.7%.

Here is the summary with the relative percentage for owner occupied housing and personal finance rising (so putting more pressure on household debt ratios in a flat income, rising cost market). Overall lending to business, relative to all lending fell again.

Personal credit is rising, now, as households find their cash flow is under pressure, many are now seeking fixed loans to help bridge the gap left by falling savings.  In prior years there was a fall at this time of year, before the Christmas binge, but that is different this year. This does not bode well for Christmas spending, and we see signs of the New Year sales already underway!

Then finally, if we look at the fixed business lending, and split it into lending for property investment and other business lending, the horrible truth is that even with all the investment lending tightening, relatively the proportion for this purpose grew, while fixed business lending as a proportion of all lending fell.

These a clear signs of a sick economy (in the sense of unwell!), with business investment still sluggish, still too much lending on property investment, and as we showed above, too much additional debt pressure on households.

I will repeat. Lending growth for housing which is running at three times income and cpi is simply not sustainable. Households will continue to drift deeper into debt, at these ultra low interest rates. This makes the RBA’s job of normalising rates even harder.

The mid-year economic forecast, later in the week will likely simply underscore the fact the economic settings are not appropriate. And, by the way, tax cuts, even if they could be paid for, will not help.

Sydney Leads Home Prices Lower

Further evidence of a fall in home prices in Sydney, as lending restrictions begin to bite, and property investors lose confidence in never-ending growth. So now the question becomes – is this a temporary fall, or does it mark the start of something more sustained? Frankly, I can give you reasons for further falls, but it is hard to argue for any improvement anytime soon.  Melbourne momentum is also weakening, but is about 6 months behind Sydney.

The Residential Property Price Index (RPPI) for Sydney fell 1.4 per cent in the September quarter 2017 following positive growth over the last five quarters, according to figures released today by the Australian Bureau of Statistics (ABS).

Sydney established house prices fell 1.3 per cent and attached dwellings prices fell 1.4 per cent in the September quarter 2017.

Hobart leads the annual growth rates (13.8%), from a lower base, followed by Melbourne (13.2%) and Sydney (9.4%). Darwin dropped 6.3% and Perth 2.4%.

“The fall in Sydney property prices this quarter was consistent with market indicators,” Chief Economist for the ABS, Bruce Hockman said.

Falls in the RPPI were also seen in Perth (-1.0 per cent), Darwin (-2.6 per cent) and Canberra (-0.2 per cent). These were offset by rises in Melbourne (+1.1 per cent), Brisbane (+0.7 per cent), Adelaide (+0.7 per cent) and Hobart (+3.4 per cent).

For the weighted average of the eight capital cities, the RPPI fell 0.2 per cent in the September quarter 2017. This was the first fall in the RPPI since the March quarter 2016.

“Residential property prices have continued to moderate across most capital cities this quarter,” Mr Hockman said.

The total value of Australia’s 10.0 million residential dwellings increased $14.8 billion to $6.8 trillion. The mean price of dwellings in Australia fell by $1,200 over the quarter to $681,100.

First Time Buyers Keep The Property Ship Afloat [For Now]

The ABS released their housing finance data to October 2017 yesterday.

Overall lending was pretty flat, but first time buyers lifted in response to the increased incentives in some states, by  4.5% in original terms to 10,061 new loans nationally.

At a state level, FTB’s accounted for a 19% per cent share in Victoria and 13.7% in New South Wales, where in both states, more favourable stamp duty regime and enhanced grants were introduced this year. But, other states showed a higher FTB share, with NT at 24.8%,  WA at 24.6%, ACT at 20.1% and QLD 19.7%. SA stood at 13% and TAS at 13.3%.

There was a shift upward shift in the relative numbers of first time buyers compared with other buyers (17.6% compared with 17.4% last month) , still small beer compared with the record 31.4% in 2009. These are original numbers, so they move around each month.

The number of first time buyer property investors slipped a little, using data from our household surveys, down 0.8% this past month. Together with the OO lift, but total first time buyer participation has helped support the market.

Looking across the data, the trend estimate for the total value of dwelling finance commitments excluding alterations and additions fell 0.3%. Owner occupied housing commitments fell 0.1% and investment housing commitments fell 0.5%. However, in seasonally adjusted terms, the total value of dwelling finance commitments excluding alterations and additions rose 0.6%.

The monthly flows show that owner occupied lending fell $23m compared with the previous month, down 0.15%, while investment lending flows fell 0.5%, down $60m in trend terms. Refinanced loans slipped 0.13% down $7.5 million. The proportion of loans excluding refinanced loans for investment purposes slipped from a recent high of 53.4% in January 2015, down to 44.6% (so investment property lending is far from dead!)

Here is the breakout by category.

We see that from a trend monthly perspective, only secured finance for owner occupied purchase of new dwellings, and construction for rent rose.

In trend terms, the number of commitments for the purchase of new dwellings rose 1.0% and the number of commitments for the purchase of established dwellings rose 0.3% while the number of commitments for the construction of dwellings fell 0.5%.

The stock data shows the value of all loans rose 0.49% or $7.8 billion (still an annual equivalent rate of three times income or inflation). Investor stock was 34.5% of all loans, slightly down from last month, but still a substantial proportion of the total.


The stock data (in original terms) showed a 0.67% rise in owner occupied loans worth around $7.1 billion and investor loans rose by 0.14% of $764 million.

So, overall the market is being supported by first time buyers, and some refinancing, reflecting the attractor rates currently on offer and recent incentives. But the fact is overall housing debts are rising, creating problems later as household debt rises, relative to income.

Worth also highlighting that many will not see their property lift in value, if the trends in Sydney continue and spread. So many first time buyers are coming in close to the top and when wages are static. So it is important to allow sufficient capacity to handle these risks and that underwriting standards are adjusted accordingly.   Trends here continue to mirror events in the USA in 2005/6. Caveat Emptor!

Wages Growth, Under The Skin, Is Concerning

The Treasury published a 66-page report late on Friday – “Analysis of Wage Growth“.

It paints a gloomy story, wage growth is low, across all regions and sectors of the economy, subdued wage growth has been experienced by the majority of employees, regardless of income or occupation, and this mirrors similar developments in other developed western economies. Whilst the underlying causes are far from clear, it looks like a set of structural issues are driving this outcome, which means we probably cannot expect a return to “more normal” conditions anytime some. This despite Treasury forecasts of higher wage growth later (in line with many other countries).

We think this has profound implications for economic growth, tax take, household finances and even mortgage underwriting standards, which all need to be adjusted to this low income growth world.

Here are some of the salient points from the report:

On a variety of measures, wage growth is low. Regional mining areas have experienced faster wage growth, but wage growth has slowed in both mining and non-mining regions. Wage growth has been fairly similar across capital cities and regional areas, although the level of wages is higher in the capital cities.

The key driver of wage growth over the long-term is productivity and inflation expectations. This means that real wage growth – wage growth relative to the increase in prices in the economy – reflects labour productivity growth. However, fluctuations across the business cycle can result in real wage growth diverging from productivity growth. There are two ways of measuring real wages. One is from the producer perspective and the other is from the consumer perspective. Producers are concerned with how their labour costs compare to the price of their outputs.

Consumers are concerned with how their wages compare with the cost of goods and services they purchase.

Generally, consumer and producer prices would be expected to grow together in the long-term, so the real producer wage and real consumer wage would also grow together. Consumer and producer prices diverged during the mining investment boom due to strong rises in commodity export prices. The unwinding of the mining investment boom and spare capacity in the labour market are important cyclical factors that are currently weighing on wage growth.

It is unclear whether these cyclical factors can explain all of the weakness in wage growth. Many advanced economies are also experiencing subdued wage growth. In particular, labour productivity growth has slowed in many economies. However, weaker labour productivity growth seems unlikely to be a cause of the current period of slow wage growth in Australia. Over the past five years, labour productivity in Australia has grown at around its 30-year average annual growth rate.

Wage growth is weaker than the unemployment rate implies. There may be more spare capacity than implied by the employment rate. [Is The Phillips curve broken?]. Labour market flexibility is a possible explanation for the change in the relationship between wage growth and unemployment, and the rise in the underemployment rate. Employers may be increasingly able to reduce hours of work, rather than reducing the number of employees when faced with adverse conditions. This may be reflected in elevated underemployment.

It is difficult to draw firm conclusions on the effect of structural factors on wage growth, given they have been occurring over a long timeframe and global low-wage growth is a more recent phenomenon. Three key trends are the increasing rates of part-time employment, growth in employment in the services industries, and a gradual decline in the share of routine jobs, both manual and cognitive, and a corresponding rise in non-routine jobs.
Both cyclical and structural factors can affect growth in real producer wages and labour productivity, so such factors can also affect the labour share of income. Changes in the labour share of income occur as a result of relative growth in the real producer wage and labour productivity. Since the early 1990s, the labour share of income has remained fairly stable. Nonetheless, different factors have placed both upward and downward pressure on the labour share of income.

An examination of wage growth by employee characteristics using the Household Income and Labour Dynamics in Australia (HILDA) survey and administrative taxation data suggests that recent subdued wage growth has been experienced by the majority of employees, regardless of income or occupation. Workers with a university education had higher wage growth than those with no post-school education over the period 2005-2010, but have since experienced lower wage growth than individuals with no post-school education.

An examination of wage growth by business characteristics using the Business Longitudinal Analysis Data Environment (BLADE) suggests that higher-productivity businesses pay higher real wages and employees at these businesses have also experienced higher real wage growth. Larger businesses (measured by turnover) tend to be more productive, pay higher real wages and have higher real wage growth. Capital per worker appears to be a key in differences in labour productivity and hence real wages between businesses, with more productive businesses having higher capital per worker.

Wage growth is low across all methods of pay setting. In recent years, increases in award wages have generally been larger than the overall increase in the Wage Price Index. At the same time, award reliance has increased in some industries while the coverage of collective agreements has fallen. There are a range of reasons for the decline in bargaining including the reclassification of some professions, the technical nature of bargaining, natural maturation of the system and award modernisation which has made compliance with the award system easier than before.

GDP Sags Below Expectation In September 2017

The ABS released the National Accounts to September 2017 today.  The expectation was a 0.7% lift in GDP, but it came in 0.6%. This gives an annual read of 2.3%, well short of the hoped for 3%+. Seasonally adjusted,  growth was 2.8%. Business investment apart, this is a weak and concerning result.  The terms of trade fell.

The GDP per capita and net disposable income per capita both fell, which highlights the basic problem the economy faces.  The dollar fell on the news.

Actually, we need to reboot our thinking on economic progress, as a quest for continual growth on the current settings will lead us into the gutter. Time for some fresh ideas. But then it seems that the alignment between potential growth, and lifts in tax take which follow makes this difficult.

The Australian economy grew 0.6 per cent in seasonally adjusted chain volume terms in the September quarter 2017, according to figures released by the Australian Bureau of Statistics (ABS) today.

Chief Economist for the ABS, Bruce Hockman, said: “Increased activity in both private business investment and public infrastructure underpinned broad growth across the industries.”

Compensation of employees (COE) increased in all states and territories, resulting in a national quarterly growth of 1.2 per cent and growth of 3.0 per cent since the September quarter 2016. Despite higher household income, household consumption was weak at 0.1 per cent, in line with the retail trade estimates. This weak household spending combined with growth in household income resulted in an increase in the household saving ratio for the first time in five quarters.

Mr Hockman added: “The increase in wages was consistent with the stronger employment and hours worked data that has been reported in the labour force survey.”

Net exports contribution to growth was flat this quarter despite higher Mining production and exports of coal and iron ore. The terms of trade fell 0.4 per cent on the back of lower export prices.

17 out of 20 industries recorded positive growth this quarter driven by Professional, Scientific and Technical Services, Health Care and Social Assistance and Manufacturing. A longer-term analysis of the changing drivers of the economy, from an industry perspective, is provided in a feature article included in this quarter’s publication.

Trend Retail Turnover Still Stagnant In October

According to the ABS, the trend estimate for Australian retail turnover fell 0.1 per cent in October 2017 following a relatively unchanged estimate (0.0 per cent) in September 2017. Compared to October 2016 the trend estimate rose 1.8 per cent. Trend estimates smooth the statistical noise.

Food retailing rose 0.1% in trend terms, Housing good retailing fell 0.6%, Clothing and footwear fell 0.1%, Department stores rose 0.2% and cafes and food services rose 0.1%.

Looking across the states, again in trend terms, NSW and VIC both fell 0.1%, WA fell 0.3%, NT fell 0.4% and ACT fell 0.2%. SA was the only state to register a rise, of 0.1%.

Online retail turnover contributed 4.7 per cent to total retail turnover in original terms.

For the record, Australian retail turnover rose 0.5 per cent in October 2017, seasonally adjusted, and follows a 0.1 per cent rise in September 2017.

In seasonally adjusted terms, all states rose. There were rises in Victoria (1.0 per cent), New South Wales (0.3 per cent), South Australia (1.2 per cent), Western Australia (0.5 per cent), Queensland (0.1 per cent), the Northern Territory (1.7 per cent), the Australian Capital Territory (0.6 per cent) and Tasmania (0.5 per cent).

Australian Debt Servicing Ratios Higher and More Risky

The Bank for International Settlements released their updated Debt Service Ratio (DSR) Benchmarks overnight. A high DSR has a strong negative impact on consumption and investment.

Australia (the yellow dashed line) is second highest after the Netherlands. We are above Norway and Denmark, and the trajectory continues higher. Further evidence that current regulatory settings in Australia are not correct. As the BIS said yesterday, such high debt is a significant structural risk to future prosperity.

The DSR reflects the share of income used to service debt and has been found to provide important information about financial-real interactions. For one, the DSR is a reliable early warning indicator for systemic banking crises.

The DSRs are constructed based primarily on data from the national accounts. The BIS publishes estimated debt service ratios (DSRs) for the household, the non-financial corporate and the total private non-financial sector (PNFS) using standardised data inputs for 17 countries.