The Anglican Church, once a key institution in the English-speaking world, has suffered decline for over half a century. Although in both the UK and North America there are many examples of growing and lively Anglican churches, as national denominations, the trend is downwards. This decline is in marked contrast to continued Anglican growth in Africa and other parts of the world. There the church is healthy. In the West, it is sick. The question is – is the Anglican sickness unto death?  

In this blog, I explore the different patterns of Anglican decline through four denominations: the Church of England (C of E), the Church in Wales (C in W), the Scottish Episcopal Church (SEC), and the Episcopal Church of the USA (ECUSA). The study is not perfect, nor is the data, but I hope it inspires debate and other studies. A subsequent blog will suggest possible reasons for their differences in decline.

1. The Pattern of Decline

Which of the four denominations is the healthiest, and which has the worst decline?  

First, look at the attendance data since the beginning of the century. I have used a short period as churches tend to change their method of measuring attendance over time, which will skew any predictions. A shorter time frame will help reduce this effect.  The attendance data is graphed in figure 1 [1][2].

Figure 1: Attendance of C of E, ECUSA – left scale; C in W – right scale.

It is clear the Church of England is the largest of the 3 denominations. Indeed, it is larger than ECUSA (left scale) even though the USA is over six times the size of England.  Numerically ECUSA has never had the position in the USA that the Church of England has had in England. Nevertheless, it could still command influence on US society, perhaps because it inherited the C of E’s prestige.  By contrast, the C in W is much smaller (right scale), reflecting its place in a much smaller country.

Rate of Decline

Bigger differences emerge when the rate of decline is examined. One measure of decline is the slope of the line through the data points. Here the C of E has the slowest decline, ECUSA has the fastest, and the C in W is somewhere in the middle. Clearly, the Church of England is healthier than the other two.

When viewed as percentages, ECUSA had a 2.7% per annum decline in 2010, whereas the C in W had 2.9%. So why is ECUSA declining faster? Simply because it is much larger. 2.7% of a big number is a big number! Percentages are misleading as the above declines are not exponential but largely linear, as ageing is part of the process. As such, the percentage decline of all will increase in time.  In 2010 the C of E had an annual decline of 1.1%, which means it is losing less in absolute terms than ECUSA. Thus ECUSA is the worst declining of the three.  

2. Extinction

How likely is it that these denominations will become extinct if current trends continue?  

Attendance and membership data for all four denominations are fitted to the limited enthusiasm model of church growth, a model that is able to use data and decide whether a declining church is heading for extinction or not [2] [3].

The Church in Wales, the Scottish Episcopal Church and the Episcopal Church of the USA are all firmly under the extinction threshold. By this, I mean that for the range of model parameters that calibrate to the data, then all resulting simulations indicate future extinction. By contrast, the Church of England is on the margins of extinction. Some calibrations say yes, just. Some say no, just [4]. Again there is a clear distinction between the C of E and the others.  

Extinction Dates?

The limited enthusiasm model was not set up to predict when extinction may occur [5], only that it will. To estimate an extinction date, linear extrapolation is used on the basis of the recent attendance and membership data, as they are approximately straight, noted previously. The results are displayed in table 1, with the graph of the attendance results in figure 2 (membership used for the SEC [2].)

Expected Extinction DateChurch of EnglandECUSAChurch in WalesScottish Episcopal Church
Table 1: Predicted Extinction Dates Using Linear Regression and extrapolation (note *).
Figure 2: Projected Attendance for C of E, ECUSA – left scale; C in W, SEC – right scale

ECUSA, C in W and SEC attendance figures all predict extinction dates around 2040 [6] figure 2. This date is confirmed for the latter two by the projected membership data, with the ECUSA membership predicting an extra 15 years, table 1. Membership data for Anglican churches tend to be unreliable as it relies on electoral roles that are only maintained periodically and have fairly minimal criteria for inclusion. Thus I would go with the figures predicted by attendance.

By contrast, the Church of England’s extinction is at the end of the century, so far away that it effectively says it is not clear if its decline results in extinction or not. Again there is a clear difference between the C of E and the other three denominations.

There is a certain amount of “wiggle” room in all these results, but not enough to delay the extinction of the denominations by much. If current trends continue, the Episcopal churches of the USA, Scotland and Wales are near the end of their lifespans and will see massive church closures from around 2025 onwards.

3. Long-Term Patterns

Where does this church decline sit in the broad scheme of the churches’ histories?  

There are no reliable attendance figures going back through the 20th century. Instead, membership figures are used, taken as a percentage of the population of each country. This will allow for population growth. The results are compared, on the same scale, in figure 3, from 1900.

Figure 3: Membership as a percentage of the population for C of E, ECUSA, C in W, and SEC.

In the past, both the Church of England and the Church in Wales have had a greater share of their national populations than that of either SEC or ECUSA. This could reflect the fact that they were the “conformist” traditions in their lands, unlike Scottish Episcopalians, who were non-conformists among Presbyterians, and ECUSA, who were merely one of many denominations. It may also reflect considerable over-reporting of electoral roles by the established churches earlier in the century. The steeper decline in the C of E from 1970 probably represents a better definition of membership coming into use! It has stabilised in this century.  

By 2000, the C of E and C in W had similar membership percentages, despite differing attendance percentages, 2.4% compared with 1.6%, respectively. It is likely the C in W, with almost double the number of “members” compared with attenders, has much over-reporting on its electoral roles.


Thus, generally speaking, the Church of England commands more loyalty among society than ECUSA, Scottish Episcopal Church or the Church in Wales. Its decline is slower, and it is unlikely to face extinction this century, unlike the other three, which have 25-35 years remaining.  Given the likely acceleration of church closures that will start in the next decade, these three Anglican denominations probably have less than 10 years to address the issue of their impending extinction.

I should also note that none of the four denominations has ever commanded widespread public loyalty in terms of membership or attendance. Churches in the West have never been as popular as they have perceived themselves to be. The church might find the future easier to face by keeping in mind its mission and its Lord rather than some idealistic picture of a past golden age that never really existed.  

A subsequent blog will explore some reasons for differences in decline between these denominations.


[1]  Data Sources  

Church of England
Statistics for Mission 2012, (2014), Archbishops’ Council, Research and Statistics, Central Secretariat. The 2012 and previous versions are no longer available. Most data is contained in later versions, e.g. 2014.

Church in Wales
Church in Wales Membership and Finances, 2013: and previous editions (no longer online).
The Arthur Rank Centre (no longer online).
Wales Online:
Wales is no longer a nation of churchgoers but faith is alive.
Church in Wales reacts to shortage of vicars

Scottish Episcopal Church
The Church Times:
Scots need greater numbers ‘to pay the rent’

Episcopal Church of the USA
The Association of Religious Data Archives:
Episcopal Church of the USA website. The data is no longer available.

Other UK membership data: Various issues of UK Church Statistics and Religious Trends, Peter Brierley, Brierley Consultancy and Christian Research.

[2] There was too little attendance data to use for the Scottish Episcopal Church.

[3] I developed The limited enthusiasm model to describe the dynamics of revival, see
A General Model of Church Growth and Decline, Journal of Mathematical Sociology, 29(3), 177-207, 2005.

When modelling decline, it is possible to cut the model down and aggregate leaving rates with birth and death rates making model calibration simpler yet still giving the same results on extinction thresholds.  

[4]  See two previous blogs on the Church of England:
The Decline of the Church of England and the update.

5] The limited enthusiasm model does not deal with age categories and assumes constant death rates. In the last few years of a church’s life, ageing dominates, death rates rise, and decline is faster. However, the model’s prediction of the threshold of extinction, and church numbers throughout most of history, is accurate.  

[6] Official attendance data for the Scottish Episcopal Church is too limited to draw a clear conclusion on an extinction date. Nevertheless, a date is provided by its membership figures.

Additional Note

[*] This note added after writing. It always interesting to read what others make of your work. Models need to be critiqued and become healthy for it. Two external comments worthy of note

Comment 1

I found this comment at but if has now been removed.

One huge red flag: it makes absolutely no sense to say that the C of E will become extinct in terms of membership 18 years before people stop attending. Much more likely to happen the other way round. That fact alone calls into question whether this “analyst” has any idea what he is talking about” A great comment.

Firstly. I computed the extinction data for the C of E with the two sets of data to show that the extinction date predicted by either was so far in the future that it confirmed the Limited Enthusiasm model’s prediction of “margin of extinction”. That is, it is too close to call. 10-15 years of data cannot predict dates 50+ years in the future by any method, so their difference is of no significance. All that can be said is that, if the actions that affect the C of E’s decline continue the same way, then extinction this century is unlikely.

Secondly. The C of E has been revising its method of measuring attendance by electoral role, making it more realistic. As such, it may appear artificially declining faster as revision takes effect. Eventually, this effect will fade, and membership and attendance will be closer.

Thirdly. Membership tends to lag attendance as people take time before they join a church, and there can be a delay before departed members are taken off the role. Thus, membership tends to be higher than attendance in declining churches. It is the other way around in growing ones. There can also be an age difference, with the membership in a declining church having an older profile than attendance. Young people are more likely to attend before settling on membership; older people stop attending due to infirmity but remain members. This would give the membership an artificially higher loss rate.

Fourthly. When the numbers in the church get low, then the type of “deterministic” model used here no longer works too well. Deterministic models give exact numbers at any given time and are reasonably accurate for large numbers due to averaging. But with small numbers, averages get unreliable, and modellers prefer “stochastic” models and deal with probabilities. For example, the 2039 extinction date for the Church in Wales is an average figure. When numbers get small, all you can give are probabilities that it will be 2039, 2042, 2036 etc. There will be a probability it is 2050 – but it will be a very small probability.

I realise the Christian Today review of this blog, on which the comment was based, called me an “analyst”. It sounds a bit like an expert in finance trying to gain some legitimacy for their views. I prefer to be called a mathematician – less pretentious! Scientific work should be judged on its own merits, not on the author’s description.

Comment 2

Related to this, a comment was made at Psephizo
When will the C of E be extinct?

I find the idea that you can extrapolate church attendance figures decades into the future with a simple linear model, well … is this a joke? Even if you wanted to extrapolate decades into the future (which strikes me as way beyond the realms of sanity), the simplest imaginable model would be logarithmic – assuming the church halves in size every N years. A linear model makes no sense at all.

I would completely agree, but ….

Strictly speaking, decline through people leaving and deaths should be negative exponential thus slowing down the decline. This is called a first-order balancing loop in system dynamics. This is precisely what the Limited Enthusiasm Model of church growth predicts, the model I used to decide which side of the extinction threshold each church fell. However, the data is NOT negative exponential but linear. The reason many of the people who handle church statistics know is that the church is also ageing. The leaving rate and death rates exert forces that slow the decline, but ageing exerts a force that counterbalances it. The net result is almost linear. Sadly it shows ageing, death and leaving have far more effects on the church than retaining children or conversion.

In an ideal world, I would build ageing into the Limited Enthusiasm model and let that model predict the extinction date rather than the straight-line fit. Unfortunately, mixing ageing, a discrete-time process as far as measured ages are concerned, with the social forces of the model, a continuous time process, is a notoriously hard maths problem. It is still an area of research, and there was a paper at this year’s International System Dynamics Conference on this. Additionally, the age profile over time is not that well known. So the data does not exist to calibrate that model.

Sorry, the reply is technical, but the issues are technical!


Tags: , , , , , , , ,