January 2021 – Chris Mallin Macro Perspectives

“The yawning gap”

And its implications for EA investors

The key chart

**PSC growth (YoY, rolling) minus M3 growth (YoY, rolling) since 1998** (Source: ECB; CMMP)

The key message

In typical cycles, monetary aggregates and their key counterparts move together. Money supply indicates how much money is available for use by the private sector. Private sector credit indicates how much the private sector is borrowing. Today’s ECB data release reinforces the extent to which money and credit cycles in the euro area diverged during 2020 and how atypical the current relationship between them has become.

This yawning gap has important implications for investors, especially those arguing for asset allocation shifts towards cyclical and value plays and shorter duration trades. The message from the money sector with elevated uncertainty, low confidence, weak consumption and subdued credit demand, remains a clear, “not so fast.”

Money sitting in overnight deposits drives neither GDP nor higher inflation.

The six charts that also matter

Broad money (M3) grew 12.3% YoY in December 2020, up from 11.0% in November, the fastest rate of growth since November 2007. In comparison, adjusted loans to the private sector grew by only 4.7% YoY, unchanged from November and the average growth recorded over the 2H20. The gap between the growth in money supply and private sector borrowing hit a record high of 7.6ppt (see key chart above).

**Growth in broad money (%YoY) and contribution from narrow money (ppt)** (Source: ECB; CMMP)

What is driving M3 growth? Narrow money (M1) grew 15.6% YoY, up from 14.5% in November, and contributed 10.7ppt to overall money growth (see chart above). Both the growth rate in M1 and its contribution were new, record highs. The private sector continued to increase holdings of the most liquid assets in 2020 despite earning negative real returns on those investments.

Monthly flows into household deposits since January 2019 – horizontal line indicates 2019 average flow (Source: ECB; CMMP)

Monthly deposit flows by households and corporates peaked in March-April 2020 but remain well in excess of average 2019 monthly flows. December’s monthly flow of household deposits (€53bn), for example, was 1.6x the average monthly flow recorded in 2019 (€33bn). In the 2H2020, the monthly flow of household deposits averaged €50bn. In other words, while HH uncertainty peaked in March-April 2020, it ended the year at a very elevated level (see chart above). Over the same period, the more volatile NFC deposit flows averaged €25bn, more than double the average flows recorded in 2019 (not shown here).

**Counterparts of M3 – contribution in ppt** (Source: ECB; CMMP )

From a counterparts perspective, credit to general government and to the private sector contributed 8.1ppt and 5.6ppt to money growth respectively (see chart above). For reference, the respective contributions to the 4.9% money growth in 2019 were 3.7ppt and -0.7ppt respectively. Note the important role played by credit to general government here.

**Trends in growth rates (%YoY) for NFC credit, mortgages and consumer credit** (Source: ECB; CMMP)

As before, relatively robust NFC credit (7.0%) and resilient mortgage demand (4.7%) offset weakness in consumer credit (-1.6%). Note that, EA households paid down consumer credit in three of the last four months on 2020 (see chart below).

**Monthly consumer credit flows (LHS) and YoY growth rate (RHS) for the EA** (Source: ECB; CMMP)

Conclusion

Money growth in 2020 reflected fiscal and monetary easing in response to weak private sector demand and rising savings (with added uncertainty regarding the extent to which rising savings are forces or precautionary). These trends are in direct contrast to the pre-GFC period when money expansion was driven primarily by strong, or excess, private sector credit.

**A favourite chart – different drivers and implications of M3 growth** (Source: ECB; CMMP)

At the start of 2021, some were arguing that rising money supply suggests higher inflation and supports asset allocation shifts towards cyclical and value plays and shorter duration trades. The message from the money sector with elevated uncertainty, low confidence, weak consumption and subdued credit demand, remains a clear, “not so fast.”

Please note that the summary comments above are extracts from more detailed analysis that is available separately. A more comprehensive treatment of the components and counterparts of M3 can also be found in “Don’t confuse the messages” posted in October last year.

“Fuelling the FIRE, Part IIb”

FIRE- vs COCO-based lending – a cross-EA comparison

The key chart

The balance between FIRE- and COCO-based lending in the six largest EA banking markets (Source: ECB; CMMP)

The key message

In “Fuelling the Fire, Part II”, I stressed the importance of distinguishing between different forms of credit. I made the contrast between productive, “COCO-based” credit and less-productive, “FIRE-based” credit, highlighted the shift towards greater levels of FIRE-based lending in the euro area (EA), and noted the negative implications of this trend for leverage, growth, financial stability and income inequality in the EA.

In this short, follow-on post, I add details on how the balance between these forms of credit differs between the largest EA banking sectors (NL/BE vs ES/IT) and across the EA as a whole. For interest, I also note an adaptation of Hyman Minsky’s hypothesis that states that over the course of a long financial cycle, there will be a shift towards riskier and more speculative sectors and discuss its application to EA banking briefly.

FIRE-based versus COCO-based lending across the EA

The balance between FIRE-based and COCO-based lending varies across the EA. Among the six largest banking sectors that account for just under 90% of total EA credit, FIRE-based lending ranges from 64% of total credit in the Netherlands to 43% in Italy (see the key chart above). Across the 19 EA economies, the low end of this range extends down to 40% in Slovakia and Greece, and Ireland joins the Netherlands and Belgium with a relatively high share of FIRE-based lending (see chart below).

**The wider picture across the 19 EA member states** (Source: ECB; CMMP )

Interestingly, in the Netherlands and Belgium, the two large economies with the highest share of less-productive FIRE-based lending, the NFC debt ratios are both 159% of GDP, well above the BIS threshold level of 90%. In both cases, the NFC sectors are deleveraging with debt ratios falling from peaks of 179% of GDP in the Netherlands (1Q15) and 171% of GDP in Belgium (2Q16).

**Trends in NFC debt ratios (%GDP) in the Netherlands and Belgium** (Source: BIS; CMMP)

The HH dynamics are very different however, re-enforcing the message that the EA money sectors are far from a homogenous group! In the Netherlands, for example, the HH sector has also been deleveraging since the 3Q201 when the debt ratio hit 121%. In contrast, HH leverage is increasing in Belgium, with the debt ratio hitting a new high (albeit a relatively low one) of 65% in 2Q20.

**Trends in HH debt ratios (% GDP) in the Netherlands and Belgium** (Source: BIS; CMMP)

The high levels of HH and NFC debt in the Netherlands has stimulated much research. Dirk Bezemer at the University of Groningen, for example, has studied the impact of these trends and the extent to which the financial sector helps, hurts or hinders the wider economy. His research builds on Hyman Minksy’s theory and the idea that over the course of a long financial cycle, investors will shift towards riskier and more speculative investments.

The 2008 crisis demonstrated that the Netherlands behaved in line with Minsky’s insights. It proved very vulnerable indeed to financial shocks. The country also experienced a stagnation in economic growth which was unusually long in international comparison.
Dirk Bezemer, “Why Dutch debt tells us economic growth may be fragile” 2017

Bezemer argues that Minksy’s theory can be applied to data on credit trends with “Minsky’s shift” being reflected in the decline in bank credit to the real sector (COCO-based credit) and an increase in funds flowing towards property and financial asset markets (FIRE-based credit). There are many similarities between Bezemer’s arguments and the trends highlighted in CMMP analysis. Please contact me for details.

CAGR in credit versus CAGR in nominal GDP (rolling, three years) in France (Source: BIS; CMMP)

That said, there are always exceptions to the rule. In France, for example, FIRE-based and COCO-based lending are more balanced (52%:48%) with the later growing strongly despite the fact that the NFC debt ratio hit a new high of 167% of GDP in the 2Q20.(This is the highest NFC debt ratio in this sample.) Over the past three years, the CAGR in HH and NFC credit in France has exceeded the CAGR in GDP by 4.4ppt and 5.4ppt respectively (see chart above), highlighting the fact that banking risk comes in many, different forms…

Please note that the summary comments and charts above are extracts from more detailed analysis that is available separately.

“Fuelling the Fire, Part II”

Addressing a major weakness in the current debate about debt

The key chart

**Trends in COCO- and FIRE-based lending (EUR bn, % total) since 2005** (Source: ECB; CMMP)

The key message

A major weakness in the current debate about debt in the “post-Covid world” is the failure to distinguish adequately between different forms of credit.

CMMP analysis, in contrast, draws a clear distinction between productive, “COCO-based” credit and less-productive, “FIRE-based” credit (see “Fuelling the FIRE, the hidden risk in QE”). This enabes a more accurate critique of current macro policy and a better understanding of the implications of unorthodox monetary policy (QE).

The shift towards greater levels of FIRE-based lending pre-dates the introduction of QE, but it was not until July 2016 that this form of credit exceeded COCO-based lending for the first time in the EA. The latest ECB data shows that FIRE-based lending now accounts for 52% of total lending (November 2020) reinforcing my September 2019 message that the “hidden risk in QE is that the ECB is ‘fuelling the fire’ with negative implications for leverage, growth, financial stability and income inequality in the EA.”

COCO- versus FIRE-based lending – the key concepts

In the broadest sense, lending can be spilt into two distinct types: lending to support productive enterprise; and lending to finance the sale and purchase of existing assets. The former includes lending to NFCs and HH consumer credit, referred collectively here as “COCO” credit (COrporate and COnsumer). The latter includes loans to non-bank financial institutions (NBFIs) and HH mortgage or real estate debt, referred to collectively as “FIRE” credit (FInancials and Real Estate). Dirk Bezemer (www.privatedebtproject.org) neatly distinguishes between the productivity of these different forms of lending:

COCO-based lending typically supports production and income formation
- CO: loans to NFCs are used to finance production which leads to sales revenues, wages paid, profits realised and economic expansion. Bezemer notes that these loans are used to realise future cash revenues from sales that land on the balance sheet of the borrower who can then repay the loan or safely roll it over. The key point here is that an increase in NFC debt will increase debt in the economy but it will also increase the income required to finance it
- CO: consumer debt also supports productive enterprises since it drives demand for goods and services, hence helping NFCs to generate sales, profits and wages. It differs from NFC debt to the extent that HH take on an additional liability since the debt does not generate income. Hence the consumer debt is also positive but has a slightly higher risk to stability

**Trends in outstanding stock of COCO-based loans since 2005** (Source: ECB; CMMP)

FIRE-based lending typically supports capital gains through higher asset prices
- FI: loans to NBFIs (eg, pension funds, insurance companies) are used primarily to finance transactions in financial assets rather than to produce, sell or buy “real” output. This credit may lead to an increase in the price of financial assets but does not lead (directly) to income generated in the real economy
- RE: mortgage or real estate lending is used to finance transactions in pre-existing assets rather than transactions in goods and services. Such lending typically generates asset gains as opposed to income (at least directly)

**Trends in outstanding stock of FIRE-based loans since 2005** (Source: ECB;CMMP)

Lending in any economy will involve a balance between these different forms, but the key point is that a shift from COCO-based lending to FIRE-based lending reflects different borrower motivations and different levels of risks to financial stability.

COCO- versus FIRE-based lending – the evidence from the euro area

**Split in lending between COCO-based and FIRE-based credit** (Source: ECB; CMMP)

At the end of November 2020, COCO-based lending in the EA totalled EUR 5,437bn. This is below the peak level of EUR 5,517bn recorded in January 2009. FIRE-based lending, in contrast, totalled EUR 5,792bn, 26% higher than the outstanding stock as at the end of January 2009. As can be seen in the chart above, FIRE-based lending exceeded COCO-based lending for the first time in July 2016. The current split of total loans is now 52% FIRE-based and 48% COCO-based, compared with respective shares of 45% and 55% in January 2009.

**NFC lending as a percentage of GDP since 2005** (Source: BIS; CMMP)

In other words, the shift towards increased FIRE-based lending pre-dates the introduction of QE in the EA. The shift becomes more noticeable in the post-GFC period and may also reflect the fact that NFC debt levels (expressed as a percentage of GDP) had exceeded the threshold level that the BIS considers detrimental to future growth for most of this period (see chart above).

COCO- versus FIRE-based lending – the impact of QE

Nevertheless, as noted back in September 2019, the hidden risk in QE is that the ECB is “Fuelling the FIRE” with potentially negative implications for leverage, growth, financial stability and income inequality in the Euro Area.

As noted above, while COCO-based lending increases absolute debt levels, is also increases incomes (albeit with a lag), hence overall debt levels need not rise as a consequence. In contrast, FIRE-based lending increases debt and may increase asset prices but does not increase the purchasing power of the economy as a whole. Hence, it is likely to result in high levels of leverage.

Similarly, COCO-based lending supports economic growth both by increasing the value-add from final goods and services (“output”) and an increase in profits and wages (“income”). In contrast, FIRE-based lending typically only affects GDP growth indirectly.

From a stability perspective, the returns from FIRE-based lending (investment returns, commercial and HH property prices etc) are typically more volatile that returns from COCO-based lending and may affect the solvency of lenders and borrowers.

Finally, the return from FIRE-based lending are typically concentrated in higher-income segments of the populations, with any subsequent wealth-effects increasing income inequality.

Please note that the summary comments above are abstracts from more detailed analysis that is available separately

“Out-of-synch”

What are the implications of the widening divergence between money and credit cycles?

The key chart

**The widening gap between growth in lending and growth in money supply** (Source: ECB; Bank of England; CMMP analysis)

The key message

Current money cycles in the EA and UK, (1) differ from previous cycles in terms of their drivers and implications, (2) are out-of-synch with the respective credit cycles, and (3) diverging from credit cycles at historically rapid rates. The implications here for growth, inflation, policy choices, investment returns and asset allocation are missing from many recent “2021 Investment Outlooks”.

Is it wise to ignore the “messages from the money sector”?

This week’s data releases show broad money growing at the fastest rate in the current money cycle in both the EA (11.0%) and the UK (13.9%). What is this telling us?

For some, rising money supply suggests higher inflation in the EA and the UK and supports asset allocation shifts towards cyclical and value plays and shorter duration trades. The message from the money sector with elevated uncertainty, low confidence, weak consumption and subdued credit demand is, “not so fast.”

The charts that matter

**Growth trends in EA and UK broad money aggregates in % YoY** (Source: ECB; Bank of England; CMMP analysis)

This week’s ECB and Bank of England data releases show broad money growing at the fastest rate in the current money cycle in both the EA and the UK (see chart above). M3 in the EA grew 11.0% YoY in November 2020, up from 10.5% in October. M4ex in the UK grew 13.9% YoY in November 2020, up from 13.2%.

To understand what these trends are telling us, and to understand how the current money cycle differs from previous cycles, we need to examine both the components and counterparts to broad money rather than focus simply on the headline numbers.

**Growth in M3 (% YoY) and contribution of M1 (ppt) to total growth** (Source: ECB; CMMP analysis)

From a components perspective, we can see that growth in narrow money i.e. notes and coins in circulation and, more importantly, overnight deposits is the key driver of overall money growth (see chart above). In the EA, for example, M1 grew at 14.5% YoY and contributed 9.9ppt to the 11.0% YoY growth in M3. Overnight deposits grew 15.0% YoY and contributed 8.9ppt to the total growth alone.

**Share of narrow money (M1) in broad money (M3) since 2000** (Source: ECB; Bank of England; CMMP analysis)

Recent trends are an extension/acceleration of longer-term secular shifts in the composition of EA and UK money supply (see chart above). Twenty years ago, M1 accounted for 42% and 48% of M3 in the EA and UK respectively. The shares were the same at the height of the GFC in 2008. Today, however, M1 accounts for 71% and 68% of M3 in the EA and the UK respectively. This increase in liquidity preference/money holdings reflects an increasingly lower opportunity cost of holding money as rates have fallen and, more recently, a sharp rise in both forced and precautionary savings by the regions’ households.

**Trends in monthly HH deposit flows 2019-2020 YTD** (Source: ECB; CMMP analysis)

Households are increasing their money holdings at 2-4x the average 2019 monthly rate, delaying consumption and repaying existing consumer credit with negative implications for economic growth and inflation. In the EA, household money holdings increased by EUR61bn in November, almost double the average flow of EUR33bn recorded during 2019 (see chart above). This reflects similar, albeit more volatile, trends in the UK highlighted in, “And now, the not-so-good-news” and to repeat the message in that post – money sitting in savings accounts does not contribute to GDP or higher inflation.

**Trends in monthly consumer credit flows and YoY growth rate** (Source: ECB; Bank of England)

EA households are not only delaying consumption, they are also repaying existing consumer credit. In November 2020, net repayments totalled EUR4bn, the largest amount since the peak of the first wave of the pandemic in April. On a YoY basis, consumer credit declined by 1.1%, the weakest level in the current slowdown (see chart above). In the UK, households repaid consumer credit for three consecutive months between September and November 2020 and YoY declines are the weakest since records began.

**Repeating the key chart – the widening gap between growth in lending and growth in money supply** *(Source: ECB; Bank of England; CMMP analysis)*

In terms of policy choices, monetary policy effectiveness requires stable relationships between monetary aggregates, but these are increasingly absent. The gap between growth rates in money supply and private sector credit demand is at record highs in both regions. In the EA, the gap between the supply of money (11.0%) and private sector demand for credit (4.7%) was 6.3ppt. In the UK, money supply grew 13.9%, 9.4ppt faster than private sector credit demand. This is one more reason to add to a lengthening list of why the required and sustained policy response should be, “fiscal, first and foremost.” In my next post, I will up-date my analysis to add another factor to this list i.e. credit is increasingly shifting towards less productive sectors of the economy (see also “Fuelling the FIRE – the hidden risk in QE“).

**A preview of the theme in my next post – COCO-based versus FIRE-based lending** (Source: ECB; CMMP analysis)

Conclusion

**What does this all mean?** (Source: ECB; Bank of England; CMMP analysis)

For some, rising money supply suggest higher inflation in the EA and the UK, and supports asset allocation shifts towards cyclical and value plays and shorter duration trades. The message from the money sector remains, “not so fast.”

**One of my favourite current charts! Growth in M3 and contribution of M1 and PSC.** (Source: ECB; CMMP analysis)

CMMP analysis of the components and counterparts of broad money tell a very different story from previous money cycles. The EA and UK money sectors are consistent signalling elevated uncertainty, low confidence, weak consumption and subdued credit demand – even before the introduction of further, more stringent lock-down policies in January 2021.

If there is a positive interpretation of current trends, it is that there is a large element of forced savings and hence pent-up consumer demand. That is true, but history also tells us that households and corporates can take time to adjust to major economic shocks and caution us against expecting a rapid reversal in confidence and consumption.

Please note that the summary comments above are extracts from more detailed analysis that is available separately

“And now, the not-so-good-news”

Too early for UK inflation hawks to get excitied

The key chart

**Trends in UK M4ex – is this a different form of monetary expansion?** (Source: Bank of England, CMMP analysis)

“Money sitting idle in a savings account does not contribute to GDP”
Dirk H. Ehnts, Modern Monetary Theory and European Macroeconomics

The key message

UK money supply (M4ex) increased 13.9% YoY in November 2020. Does this mean that investors positioned for a pick-up in UK inflation should be getting excited?

No, not yet (if at all).

As mentioned in previous posts, the message from the money sector is very different now from previous periods of rising money supply. The key 2020-21 messages remain:

heightened household uncertainty
weak household consumption
subdued overall credit demand

None imply ST inflationary pressures.

Household uncertainty – the chart that matters

**Monthly flows in UK households sterling money holdings in £bn** (Source: Bank of England; CMMP analysis)

Household M4 accounts for 64% of total M4ex. In November 2020, UK households increased their sterling money holdings by £17.6bn, up from £9.4bn in October. This represents the second highest monthly flow after May 2020’s £25bn and was almost 4x the size of the average monthly flow in 2019. Uncertainty reigns in the UK household sector.

Weak household consumption – the chart that matters

**Monthly flows in UK consumer credit in £bn – a negative figure indicates net repayments** (Source: Bank of England; CMMP analysis)

Households also made net repayments in consumer credit for three consecutive months between September and November 2020 of £0.8bn, £0.7bn, and £1.5bn respectively. The -6.7% YoY decline in consumer credit in November was the weakest level since the series began in 1994.

Subdued credit demand – the charts that matter

**YoY growth rates in the BoE’s headline money (M4) and credit (M4 Lending) series – out of synch!** (Source: Bank of England; CMMP analysis)

**YoY growth rate in lending minus YoY growth rate in money – abnormal money and credit cycles** (Source: Bank of England; CMMP analysis)

Finally, the gap between growth in M4 (13.9%) and ML lending (4.5%) remains at a record high of 9.4ppt. In other words, this is not a normal cycle with synchronised money and credit trends (albeit with traditional leading and lagging relationships). Consequently, and with credit demand remaining subdued, investors should be wary of assuming normal relationships between money supply and inflation (to the extent that such relationships exist at all).

Conclusion

“Monetary policy effectiveness is based on certain stable relationships between monetary aggregates.”
Richard Koo, The Holy Grail of Macroeconomics

To repeat the penultimate lesson from the money sector in 2020 – periods of monetary expansion differ in terms of their drivers and implications. The message in the pre-GFC period was one of over-confidence and excess credit demand. In contrast, the current message is one of elevated uncertainty, weak consumer demand and subdued overall credit demand (with the added uncertainty regarding the extent to which rising savings are forced or precautionary).

It is too early for UK investors who are positioned for a pick-up in inflations to get excited.

[Note that this post was drafted before the announcement of further UK lockdown restrictions on 4 January 2020]

Please note that the summary comments above are extracts from more detailed analysis that is available separately.

“First, some good news”

Positive trends for UK mortgage providers

The key chart

**Trends in UK mortgages during 2020 -balances, monthly flow and YoY growth rates** (Source: Bank of England, CMMP analysis))

The key message

The Bank of England’s latest “Money and Credit” release (4 January 2021) provides three positive data points for UK mortgage providers. First, households borrowed £5.7bn secured on their homes in November 2020, the highest level since March 2016. Second, mortgage approvals are at their highest level (105,000) since August 2017, suggesting positive future lending trends. Third, the effective rate on new mortgages increased a further 5bp in November to 1.83%, up from August’s low of 1.72%.

That said, mortgage demand remains very subdued in relation to historic trends and rates on new lending continue to act as a drag on revenues generated from outstanding mortgage balances.

As noted, back in October and December 2020, this is no time for mortgage providers to relax despite these positive developments. The winners in 2021 and beyond will be those providers who accelerate digitalisation across operations, sales and finance and risk to differentiate themselves and improve the experience for their members.

Six charts that matter

**Worth repeating the key chart again!** (Source: Bank of England; CMMP analysis)

UK households borrowed £5.7bn secured on their homes in November 2020, the highest level of monthly borrowing since March 2016. November’s monthly flow compares with average monthly borrowings of £4bn in 2019 and £2.7bn in 1H2020. The YoY growth rate in mortgages rebounded slightly to 2.9%, above the recent lows of 2.7% YoY recorded in August and October 2020.

**Looking forward – trends in approvals for house purchases since 2007** (Source: Bank of England; CMMP analysis)

Mortgage approvals in November – an indicator for future lending – hit the highest level since August 2007. Approvals for house purchases increased to 105,000 from 98,300 in October and the 2020 low of 9,349 in May. In its commentary, the Bank of England noted that, “recent strength in approvals has almost fully offset the significant weakness earlier in the year.”

**Trends in the effective rate on new mortgages** (Source: Bank of England; CMMP analysis)

The effective rate on new mortgages (the actual interest rate paid) increased a further 5bp in November to 1.83%, up from August’s low of 1.72%. This rate is, however, still down 4bp YoY and 5bp YTD. The rate on the outstanding stock of mortgages was slightly lower at 2.11% in November (a new low).

**Monthly flows (£bn) in lending to UK individuals** (Source: Bank of England, CMMP analysis)

Resilient mortgage demand is the one bright spot in an otherwise gloomy UK retail lending market. Total lending to individuals grew by only 1.6% YoY, despite the 2.9% growth in mortgage lending. The annual growth rate in consumer credit fell to -6.7% in November, another series low. Since the beginning of March, UK consumers have repaid over £17bn in consumer credit.

**UK mortgage lending since 2000** (Source: Bank of England; CMMP analysis)

That said, current mortgage demand remains very subdued in relation to past cycles (see graph above). Real YoY growth rate in mortgage has averaged only 2.2% YoY during 2020. This compares with an average real growth rate in excess of 9% YoY between November 2000 and November 2008.

**Effective interest rates on new mortgages and the outstanding stock** (Source: Bank of England; CMMP analysis)

The 28bp gap between the effective rate on new mortgage lending (1.83%) and the effective rate on the outstanding stock of mortgages (2.11%) is relatively narrow in relation to recent trends but on-going pressure on NIMs and revenue growth remains a key challenge for the sector.

Conclusion

As noted, back in October and December 2020, however, this is no time for mortgage providers to relax despite the positive developments noted above. The winners in 2021 and beyond will be those providers who accelerate digitalisation across operations, sales and finance and risk to differentiate themselves and improve the experience for their members.

Please note that the summary comments and charts above are abstracts from more detailed analysis that is available separately.