The purpose
of this research is to see if cash or shares give the better return over periods
of investment from one to twenty years.
Most
advisers would agree that over the long-term shares are the better place for
money. And most would also agree that over the short-term cash is safer. But
where is the time boundary between that advantage of shares over cash? Advisers
often call five or ten years ‘long-term’. But there can be substantial risk of
losing money over that sort of period. So where is the advantage boundary
between cash and shares?
DATA
Shares are represented by a real FTSE
100 tracker from HSBC. Chosen as a typical tracker – not cheap, not expensive,
and accurate. Data from Morningstar. The numbers are the total return with
dividends reinvested and after charges. It is assumed that once invested the
money is left untouched to the end of the period at which time it is encashed.
Cash is represented by best buy one
year bank or building society deposit accounts – referred to as ‘one-year bonds’.
The data is taken from MoneyFacts
Savings Selection and was collected from original copies of the MoneyFacts periodical which is published
each month. One year bonds are used in the analysis as what I call ‘active
cash’. On each anniversary of the investment the bond is cashed in and then
reinvested in the current best buy one-year bond. In the real world there would
be a time lag between encashing and having the money available for
reinvestment. This point is dealt with later.
All datasets
are of prices on the first of the month. Morningstar data is for the closing
price on the date. It has been normalised to begin at £10,000 in 1 January 1995.
MoneyFacts is published once a month and the data is for best buys a few days
earlier, just before the start of the month of publication. This slight date
difference is considered immaterial.
ANALYSIS OF FIVE YEAR PERIODS
The datasets
run for 21 years from 1 January 1995 to 1 January 2016. For each time period of
1 to 20 years within the dataset the growth in cash (as defined) and shares (as
defined) was calculated.
Let us take
a five-year period for example. The calculation is of the growth in the
investment over five years from an investment made on the first of a month. So
a five-year period runs for example from 1 October 1999 to 1 October 2004. Twenty-one
years of data is used from 1 January 1995 to 1 January 2016. Over that period
of 21 years there are thus 16x12=192 possible investment periods. It is assumed
that the investment is opened on day one and cashed in on the anniversary at
the end of the period.
For each of
the periods the growth is calculated for the tracker and for the cash. The two
growth values are compared to see if tracker or cash gives the higher growth
over that period. No deduction is made for tax on interest or dividends. That
approximation is explained later. For each five-year period the spreadsheet
counts the occasions when (a) cash or (b) shares gave higher growth. It also
counts those periods in which the money in the tracker ends up less than it
started.
All
calculations and data are nominal – no account is taken of inflation which
affects cash and tracker funds equally. Whatever measure of inflation was used
it would not affect whether cash or shares gave the higher return.
The results
are shown in the table
Active
cash vs FTSE 100 tracker 1/1/95-1/1/2016
|
|
5 YEAR INVESTMENT PERIODS
|
|
Cash better
|
109 periods.
57% of 192 periods
|
Shares better
|
83 periods.
43% of 192 periods
|
Shares LOSE money
|
46 periods.
24% of 192 periods
|
The table
shows that for periods starting 1 January 1995 to 1 December 2010 a five-year
investment in cash or shares (as defined) started on the first day of a
randomly chosen month would yield a better return in cash more often than one in
shares.
Investing in
shares carries another risk – losing money. In 46 out of the 192 five-year periods
(24%) the tracker would lose money, ending lower than it started. Cash never
loses money within the Financial Services Compensation Scheme limit – currently
£75,000. Above that limit there is of course a risk that the savings
institution will collapse and some or all of the capital lost. Large sums can
be protected if they are spread among several institutions keeping the money in
each at £75,000 or below. That will reduce the interest earned as by definition
not all can be in the best buy account. But the difference between the top 5 one
year bonds is not that great and would not affect the overall balance between
cash and shares. Further analysis assuming start dates throughout the month
shows the same risk of loss with shares. Any day start: 23.82% end lower. First
of month start: 23.96% end lower.
Over the 21
years from January 1995 to January 2016 the overall return for shares is 6.0%
per year compound. For active cash the overall return is 5.0% per year
compound. The 1.0% difference is significantly lower than the 3%-8% often
quoted for the so-called ‘risk premium’ of investing in shares.
Further analysis of the five-year periods
On those
occasions when shares win the gain is more than the occasions on which cash
wins. Overall for the 192 five-year periods when shares beat cash they make an
average return of 10.4% a year while cash when a winner makes just 5.5%. But of
course when shares lose they can also lose money, as the graph shows. It also
illustrates that it take a good couple of years from shares to recover from a
serious fall. These falls should not be seen as exceptional – they are quite
common.
In the 109
periods when cash beats shares investors had to wait an average of 23 months
(median 19 months) before an investment in shares would again win. And in the
46 periods when shares produced a negative return it took as much as 42 months and
as little as one period for that to be reversed. An average of 19.7 months
(median 19.5). So when shares fail to perform there may be a long wait before
investing in them is advisable again. But all that speaks to is when judgement
might be exercised in share investments. This research is not about that. It is
a comparison of using two simple rules – choose a low cost tracker vs use one
year active cash.
Tax
If the
interest on cash had been taxed at the current basic rate then over five years
the figures are closer but the balance is still in favour of cash. Shares win in
47% (91) of the 192 periods, cash wins in 53% (101). For higher rate taxpayers the
calculation is more difficult as tax would also be due on dividends which would
affect the comparison. Data to do that calculation is not available and is not
done.
Basic rate
tax is not a consideration for cash interest on ISAs or cash in pension funds as
no tax is charged on it. Note that the figures used here are not the rates for
ISAs or for funds which can be SIPPed which may be different from those used.
Such rates are not available historically over this period.
Looking
ahead from 6 April 2016 the Government estimates that 95% of cash savers do not
pay tax on the interest because the first £1000 of interest earned is tax free
(first £500 for higher rate taxpayers). So comparing untaxed cash returns with
dividends is fair for the future.
Time lag
Active cash means taking out a one-year bond and then reinvesting the money at the best buy rate when it expires. In the past there would be a delay between the end of the bond and the money being available for reinvestment. Typically that might be up to two weeks. Modelling that is difficult. One approximation is to assume interest rates are 50/52 of the recorded rate. Doing that does not affect the balance of shares vs cash which still wins in 109 out of 192 periods.
Active cash means taking out a one-year bond and then reinvesting the money at the best buy rate when it expires. In the past there would be a delay between the end of the bond and the money being available for reinvestment. Typically that might be up to two weeks. Modelling that is difficult. One approximation is to assume interest rates are 50/52 of the recorded rate. Doing that does not affect the balance of shares vs cash which still wins in 109 out of 192 periods.
Charges
The money taken by managers from a tracker fund as charges has fallen
over the years. Today the charge on the HSBC tracker used is 0.18%. In the past
it would have been more.
Summary of five year periods
Over the survey period January 1995
to January 2016, five-year investments begun on the first of each month in
active cash beat a FTSE100 shares tracker in 57% of 192 periods.
Over the survey period January 1995 to
January 2016, five-year investments begun on the first of each month in a FTSE
100 tracker ended the period lower than they started in 46 out of 192 periods (24%).
That is almost a one in four chance of losing money. Cash, within Financial
Services Compensation Scheme limits, will never lose money.
LONGER AND SHORTER TIME PERIODS
The table
below gives the results for periods from 5 to 10 years. Cash wins over 5 to 9
years with an even split over ten years.
Active cash vs FTSE 100
tracker 1/1/95-1/1/2016
|
||||||
LENGTH OF INVESTMENT PERIOD
|
||||||
WINNER
|
5 years
|
6 years
|
7 years
|
8 years
|
9 years
|
10 years
|
Shares better
|
83 periods
43% of 192 periods
|
67 periods
37% of 180 periods
|
50 periods
30% of 168 periods
|
57 periods
37% of 156 periods
|
52 periods
36% of 144 periods
|
66 periods
50% of 132 periods
|
Cash better
|
109 periods
57%
|
113 periods
63%
|
118 periods
70%
|
99 periods
63%
|
92 periods
64%
|
66 periods
50%
|
Shares LOSE money
|
46 periods
24%
|
35 periods
19%
|
13 periods
8%
|
12 periods
8%
|
14 periods
10%
|
15 periods
11%
|
Over longer
and shorter periods than that the results are surprising.
Active cash vs FTSE 100
tracker 1/1/95-1/1/2016
|
||||
LENGTH OF INVESTMENT PERIOD
|
||||
WINNER
|
1 year
|
2 years
|
3 years
|
4 years
|
Shares better
|
155 periods
65% of 240 periods
|
140 periods
61% of 228 periods
|
125 periods
58% of 216 periods
|
103
50% of 204 periods
|
Cash better
|
85 periods
35%
|
88 periods
39%
|
118 periods
70%
|
101 periods
50%
|
Shares LOSE money
|
64 periods
27%
|
64 periods
28%
|
71 periods
33%
|
60 periods
29%
|
Shares win
over 1, 2, and 3 years; cash and shares win equal times for 4 years. Over
periods longer than 10 years cash continues to dominate.
Active cash vs FTSE 100
tracker 1/1/95-1/1/2016
|
|||||
LENGTH OF INVESTMENT PERIOD
|
|||||
WINNER
|
11 years
|
12 years
|
13 years
|
14 years
|
15 years
|
Shares better
|
58 periods
48% of 120 periods
|
44 periods
41% of 108 periods
|
23 periods
24% of 96 periods
|
3 periods
4% of 84 periods
|
4 periods
6% of 72 periods
|
Cash better
|
62 periods
52%
|
64 periods
59%
|
73 periods
76%
|
81 periods
96%
|
68 periods
94%
|
Shares LOSE money
|
8 periods
7%
|
0 periods
0%
|
0 periods
0%
|
0 periods
0%
|
0 periods
0%
|
Active cash vs FTSE 100
tracker 1/1/95-1/1/2016
|
|||||
LENGTH OF INVESTMENT PERIOD
|
|||||
WINNER
|
16 years
|
17 years
|
18 years
|
19 years
|
20 years
|
Shares better
|
7 periods
12% of 60 periods
|
18 periods
38% of 48 periods
|
29 periods
81% of 36 periods
|
21 periods
88% of 24 periods
|
12 periods
100% of 12 periods
|
Cash better
|
53 periods
88%
|
30 periods
63%
|
7 periods
19%
|
3 periods
12%
|
0 periods
0%
|
Shares LOSE money
|
0 periods
0%
|
0 periods
0%
|
0 periods
0%
|
0 periods
0%
|
0 periods
0%
|
Cash wins over
investment periods from 11 to 17 years, and among them from 13 to 16 years cash
absolutely takes over. Over 14 year periods cash wins in 96% of the periods,
shares in just 4%. In periods of 18 to 20 years shares win comprehensively with
shares doing better in 100% of the twelve 20 year periods. Sample sizes get
smaller as durations lengthen and are therefore less reliable.
For investment
periods from 1 to 11 years there are always some starting dates when an
investment in shares finishes lower than it started. For periods of 1 to 4
years it is always over 25%, peaking at 33% for three year periods. For 11 year
periods it is 7% - one in 14. For investment periods from 12 years onwards the
tracker always ends higher than it started.
These tables
are depicted in the graph below.
Starting date
Further
analysis of the data shows that for many starting dates from 1 January 1995
investing in shares over any possible period from one year upwards would have
produced a lower return than using properly managed best buy cash. That is true
for example for the whole of the two years from 1 October 1999 to 1 September
2001 and for four months from 1 October 2007 to 1 January 2008. Money invested
on the first of any month on those dates and left for any period from 1 year to
the maximum possible 15 or 16 years would have done better in cash than shares.
There are
fewer comparable times when shares produced a higher return over every possible
investment period:- 1 November 2008 to 1 September 2009 is the longest run and
two earlier dates are 1 October 2002 and 1 October 2003.
Overall for
investment periods of five years or more there are 38 starting dates when cash
would always have produced a better return but only 24 starting dates when that
was true of shares.
Total winner
Adding all
the data for all investment periods from one to twenty years cash wins in 55.7%
of the 2520 periods and shares in 44.3% of them. Overall cash beat shares.
21 Years
Over the
whole 21 year period the tracker out-performs cash. £10,000 invested in a
tracker on 1/1/1995 becomes £34,098 by December 2015, a compound growth rate of
6.0% a year. Cash achieves nearly £6000 less finishing at £28,105, a compound
growth rate of 5.0%. But the low difference between the risk free cash and the
risky shares is striking. Bodie (Financial
Analysts Journal May-June 1995 pp. 18-22) finds that the risk of investing
in shares does not diminish as time invested lengthens.
CONCLUSION
The data presented here allow the boundary
between cash and shares to be set at around 18 years. Less than that there is a
better than evens risk that a shares tracker will produce a lower return than a
series of best buy cash accounts. For periods below 12 years there is also a
risk that a shares investment will lose money. Overall, for a random date and a
random investment period the safer bet is active cash rather than tracker
shares.
NOTES
Tracker choice
I wanted to
use a typical FTSE100 tracker that was not low or high cost and might have been
picked by an unskilled investor in 1995 for which data was available.
MorningStar provided the HSBC FTSE 100 Index Retail Income which fits that
description and is used for this analysis. Other trackers are available of
course. The five-year analysis has been repeated using HSBC FTSE 100 Index
Retail Accumulation, the Marks & Spencer UK [FTSE] 100 Comp Acc, the HSBC
FTSE 250 Index C Acc, and the All Share tracker M&G Index tracker A Acc.
The data are
not available over the same period and have been taken from 1 March 1998 to 1
April 2016. The results from the FTSE100 trackers are almost identical with the
data here. The FTSE All Share tracker shows a similar result with cash beating
it in 53% of five-year periods. However, the FTSE 250 tracker has the opposite
result with shares beating cash in 65% of periods. The FTSE 250 has shown
particular growth recently. However, it would not have been a choice in 1995.
Another
exercise using just the HSBC Acc and the M&S Acc FTSE 100 trackers from 1
January 1996 (the longest run of data available for the two) showed that over five-year
periods the HSBC Acc produced a 57:43 split in favour of cash and the M&S
Acc tracker showed 55:45 in favour of cash.
Of course
with hindsight there will always be trackers which do better and others which
would do worse. This research was not designed to do that study. It was designed
to compare a typical tracker which would be chosen in 1995 over subsequent five-year
periods. At that time the FTSE100 would have been the obvious choices. The
figures show that the exact tracker of this type which was used does not affect
the overall conclusions.
Linked periods
The periods used are not of course independent of one another as they
overlap. That raises statistical problems with considering the significance of
the differences. But the research was not designed to look for that. It simply
says ‘if over the last 21 years I was to stick a pin in a date for an
investment and then pick the investment period x, which would give me the
better return, active cash or passive tracker?’ Chances are over almost all
values of x it would be cash. And for all value of x<12 there is a chance of
losing money in a tracker.
It also emphasises the importance of when the series starts. That is
reflected in the fact that the periods when cash or shares win tend to come not
at random but in blocks. The overwhelming advantage of cash over 14 years
reflects the dates on which share price collapses and interest rate cuts fall.
Other studies
This research
gives different results to every other study of shares vs cash. They universally
state that shares do better than cash over every long-term period. The biggest
and best known research is published as the annual Barclays Equity Gilt Study (BEGS). Now in its 61st year
it now covers periods from 1899. The results of the latest 2016 study are
unequivocal in showing cash bottom of investment classes and, by deducting
inflation, giving negative real terms returns.
Shares: The BEGS however overstates the
return on shares. It uses its own total return index which records the growth
of UK shares with dividends reinvested. But it does not include the impact of
charges so the stated growth would not be achieved by any real investor.
Charges can cost 1% or 2% a year. Even a modern tracker or ETF makes some annual
charge which can be 0.5% or more, though many are less. Other charges for
buying and selling shares for example are not openly stated but affect returns.
Any annual charge has a profound effect over the longer term. One of the
authors of the report explained to me why charges are not included in its indexes
“trends in transaction costs may have varied significantly over time. It is
difficult to ascertain a history of a representative transaction costs.”
My research
uses a real tracker – the HSBC FTSE 100 Index Retail Inc – so represents real
returns after charges. It was chosen as a typical FTSE tracker for which data
were available over a 21 year period. They are rare.
A comparison
of the returns on the BEGS index and the tracker used in this study from 1995
to 2015 shows that the BEGS index grows 336.80% which is compound growth over
the 21 years of 7.3%. My real tracker grows 240.98% which is compound growth of
6.0%. That 1.3% advantage of the BEGS study reflects the difference between a
proprietary calculated index without charges and the real world returns on a
named FTSE100 tracker.
Cash: The BEGS understates the return
on cash. For its analyses it uses using a cash surrogate – three month Treasury
Bills and for the annual figures uses the return on four successive three-month
bills. That data shows compound growth over 21 years from 1995-2015 of 3.9%.
The 2015 return on Treasury Bills is 0.45%. In January 2016 the best buy
instant access savings account paid 1.65% and the Bank of England average over
the 12 months to 1 January 2016 was 0.47%.
For
comparison the BEGS also uses the interest paid on cash in an instant access
postal building society account. That data shows compound growth over 21 years
from 1995-2015 of 2.7%. The Nationwide account used since 1998 currently pays
interest of just 0.25% a year.
Since 1998 the
BEGS has also used a now closed Nationwide postal account – Nationwide Invest
Direct – which currently pays 0.25%. In past years the rate it has used has
been more competitive and for some years was better than the Bank of England
recorded average. But since 2008 that has not been the case. This new data show
that the rate used by BEGS has consistently been lower than the best buy rates for
instant access accounts recorded by MoneyFacts. Both are well in excess of the
BEGS rate of 0.25%. Over the 21 years the average rates are BoE 1.59%, BEGS
2.67%, MF best buys 4.51%. Clearly using the MF best buys on instant access
would make a significant difference to the returns on cash over the period. This
study uses one year bond rates as part of the Active Cash model explained
above. They are normally higher than instant access rates.
One of the
BEGS authors tells me
“We refer to
3 month Treasury Bills as “cash” for both the US and UK. We provide UK building
society data as well in order to provide an alternative measure of cash
returns. We used to follow Halifax returns but since that was converted to a
bank, we chose Nationwide as one of the last remaining large building
societies. The source for “best buys” will change over time and may come with
differing conditions such as investment amount, or holding period. Again, our
aim here is to provide a consistent long run series, a measure of cash returns
over a 100 years, so we use the Treasury bills.”
BEGS is
correct that over the long run it would be difficult to ascertain either investment
charges or real cash data. But by ignoring charges and failing to use a better
version of cash the study inevitably exaggerates the real returns on shares and
understates the returns on best buy active cash.
Past and future
In general the past in investment is no guide to the future. A fund
which has performed well in the past does not help us decide if it will perform
well in future. Indeed, costs are the key determiner of which funds do best in
the long-term – the lower the cost the better. The FSA (as was) found that poor
past performance did tend to persist, but good did not.
This study is not bound by that general investment rule. Over the 21
years it covers there has been great variety in both share price changes and
cash returns. There is no reason to believe that will be different in the
future. The safest assumption for conclusions is that the 21 year study – a
period chosen simply on the availability of data – is typical and would be
reflected in the future. That is because it is driven by analysis of objective
data which does not arise directly from human judgements. In that sense it is
more physics than investment and uses the assumption that we occupy no special
place or time. What was true will be true. At least that is the safest
assumption.
Data from the BEGS confirms that the period 1995 to 2015 is not that
different from the whole period from 1899 to 2015. For example, over the long
run there is a significant percentage of investment periods when cash
outperforms shares. The proportion identified is clearly less than the present
study due to the overstatement of share returns and the understatement of cash
returns. But for example from 1899 to 2015 30% of three year investment periods
showed a higher return for cash. From 1995 to 2015 that percentage was 33%.
Similarly cash won in 25% of investment periods of five years from 1899-2015
compared with 29% from 1995 to 2015. Those figures do not imply that the period
1995 to 2015 is anything special. Taking the available data from the table
below cash outperforms shares in 21% of periods 1899-2015 and 23% 1995-2015.
Over n years
|
||||||
2
|
3
|
4
|
5
|
10
|
18
|
|
1899-20151
|
32%
|
30%
|
27%
|
25%
|
9%
|
1%
|
1995-20152
|
24%
|
33%
|
38%
|
29%
|
14%
|
0%
|
1995-20153
|
39%
|
42%
|
50%
|
57%
|
50%
|
19%
|
Sources:
1. BEGS figure 8.1 p.61 cash =
4x3 month Treasury Bills; shares = own total return index
2. BEGS email to me 9 June 2016;
3. Paul Lewis research. Cash =
Active Cash. shares = HSBC FTSE100 tracker.
Paul Lewis
15 June 2016
