The economics of church decline in Scotland.
Studies of the economics of the church are scarce. Despite a seminal contribution from Adam Smith (see Anderson, 1988) economists have paid only cursory attention to this topic, doubtless partly because it requires the extension of economic reasoning to a non-market exchange problem. The extant literature is dominated by contributions in the Chicago-inspired household production framework which tests models using North American data. Given the significant differences in transatlantic religiosity, there is clearly warrant and scope for examining religious behaviour in the United Kingdom. In this article we study Scotland.
The well-documented stylized fact of Christian religious practice in Scotland i that of increasing disinvolvement. Membership of the national kirk, the Church of Scotland, for example, has witnessed continuous absolute decline since the late 1950s. This phenomenon has attracted some attention from sociologists though, as yet, there is no consensus on the causes of waning participation in institutional religion. The novelty of our contribution lies in the demonstration of the importance of economic and socio-economic variables in explaining the secular diminution in demand for the (Sunday) services of the Scottish church.
The first part of the article takes a closer look at time series data points fo the various Scottish church denominations. Data availability considerations constrain our focus to be that of the Church of Scotland. Analysis of its membership inflows and outflows reveals that the decline in the stock of church members is accounted for in terms of the ebbing of the former. Theoretical explanations of this observation are surveyed and used to motivate the specification of a simple demand equation. The results of the estimation are discussed and the principal conclusions of the research outlined.
Scottish Church Data
Participation data, usually measures of average attendance, are the most sensitive indicators of levels and changes in formal religious practice. Regrettably, Scottish attendance data are available only intermittently. Recent and reliable point estimates were collected in a census of church attendance conducted in March 1984 (Brierley and Macdonald, 1985), but the previous comparable national census of religious worship in Scotland dates back to that directed by Horace Mann on 30 March 1851 (Parliamentary Papers, 1854)|1~.
Given the paucity of attendance data, a time-series analysis must rely on membership figures. The membership data for the main denominations in the years 1947 and 1984 are set out in Table I.
In the case of the Roman Catholic church these are measured on a demographic basis, counting as members all those baptised into its ranks in infancy as well as those similarly initiated at a later stage of the life cycle. The increase over the period partly reflects Irish immigration during the late 1940s and the 1950s and high birth rates. Figures for the total Catholic population per se ar not terribly interesting because of the inclusion of minors and the failure to discriminate between lapsed and practising adults. Mass attendance data collate by Highet (1960 and 1985) for adult Catholics (aged 20 years and over), however provide some indication of the trends in participation. His estimates disclose that, in 1959, 63 per cent of Catholic adults attended Mass in Scotland; by 198 this proportion had declined dramatically to 35 per cent (Highet, 1985, p. 10).
Significant declines are mirrored in the data on non-Catholic adults, here loosely labelled, "Protestants". In 1947 church membership embraced 45.8 per cent of Protestants. By 1984 the share had fallen significantly to 31.9 per cent. In absolute terms, this represents an average loss of 11,475 members per year; an annual rate of decline of 0.9 per cent. At the level of aggregate formal ecclesiastical adherence, therefore, the post-war period has witnessed a spectacular slide in Christian religious practices.
To explain the net downward trend, the remainder of the article will focus on the Church of Scotland. As Table I indicates, the national kirk dominates Protestant membership, accounting for approximately 85 per cent of the total. The Roman Catholic church is omitted from the analysis due to the lack of non-demographic membership data over most of the period of interest. The other Protestant churches are overlooked on the grounds of their small size and the lack of complete membership stock and flow data in some cases.
Table I. Church Membership by Major Denomination, 1947 and 1984 Membership Denomination 1947 1984 (1) Church of Scotland 1,256,167 907,920 (2) Scottish Episcopal 55,270 39,610 (3) Other Protestant 175,571 114,910 (4) Total Protestant 1,487,008 1,062,440 (5) Total Roman Catholic population 678,538 814,400 (6) Non-Catholic adult population 3,243,983 3,325,872 (7) Protestant membership as a 45.8 31.9 percentage of (6) Sources: Highet (1950, pp. 74-75); The Catholic Directory for Scotland; Brierle and Macdonald (1985, p. 60); Registrar General for Scotland, Annual Report. Note: The non-catholic adult population was estimated by subtracting the Catholic population from the total and multiplying the result by the proportion of the Scottish population over 16 years of age.
Church of Scotland
Figure 1 plots the series for the Church of Scotland membership as a proportion of non-Catholic adults between 1950 and 1988. Note first the increase during th 1950s (continued in fact from the late 1940s) to a peak of 41.8 per cent in 1957. However, this is followed by a smooth downward trend, accentuated from 1967, such that by 1988 less than one quarter of the adult non-Catholic population were members of the Established Church. Historically, the post-war period is particularly startling due to the rapidity of decline, unprecedented both in absolute terms and growth rates.
Dissection of Inflows and Outflows
Membership flow data illumine the behaviour of the aggregate stock of members. The current stock (|M.sub.t~), defined at the end of the year, is identical by definition to the stock lagged one period plus the difference between admission (inflows, |I.sub.t~) and removals (outflows, |O.sub.t~) during the year:
|M.sub.t~ = |M.sub.t-1~ + |I.sub.t~ - |O.sub.t~
Absolute membership flows, however, are not as interesting as membership flow rates. The appropriate stock from which flows into Church of Scotland membershi arrive is the stock of adult, non-member, non-Catholics|2~; and the appropriate stock from which they leave is the stock of church members. Hence the inflow (o entry) rate is defined as the ratio of the flow into church membership during a year to the stock of non-members at the beginning of the year; and the outflow (or exit) rate as the ratio of the flow out of membership during the year to th membership stock at the beginning of the year.
Figure 2 plots the two series for the flow rates on a logarithmic scale: this captures the proportional change in the flow rates with respect to time. It suggests that the main cause of changes in the overall membership stock is in the inflow rate; the outflow rate fluctuated very little by comparison, around mean of 3.68 per cent.
Excluding intercongregational transfers, admissions are of two types: on Profession of Faith; and by resolution (for transfers from other denominations and for the restoration of lapsed members). Church of Scotland data clearly sho that admissions on Profession of Faith dominate the admissions total and these have declined precipitously. So it is the determinants of inflows in general an admissions on Profession of Faith in particular which provide the key to explaining membership declension.
The failure to maintain the recruitment rates of new members chiefly reflects the diminishing attractiveness of church membership to young people: it is this population cohort which traditionally accounts for the bulk of new entrants. Empirical evidence on conversion ages, summarized in Iannoccone (1990, p. 301) for the USA, indicates that individuals make their first serious religious commitments at a mean age of 16 or 17, typically a personal affirmation of the religion in which the subject has been reared. Conversion to a different church comes later (early 20s). Since decisions on religious commitment and, therefore church membership tend to cluster in the early part of the life cycle, the key to explaining church decline lies largely in understanding the disaffection of adolescents from the church; the failure of the inter-generational maintenance of belief.
Much of both the theoretical and applied work analyses the determinants of religious participation using an allocation of time framework (Azzi and Ehrenberg, 1975; Ehrenberg, 1977; Iannaccone, 1990; Long and Settle, 1977; Ulbrich and Wallace, 1984). Utility maximizing agents are modelled as allocatin time among market and non-market activities on the basis of perceived benefits and opportunity costs. People with a lower opportunity cost of time, as represented by market wages, are expected to display greater participation rate in a time intensive activity such as church membership. The predictions of thes models are then usually shown to be consistent with US cross-section data.
In principle, Azzi and Ehrenberg (1975, p. 36) acknowledge that the sign of the effect of higher market wages on religious participation is indeterminate a priori. Providing that church attendance is a normal good, the income effect of a rise in wages implies that a higher income can be earned with the same or les effort, leaving more time for non-market activities. In contrast, the income compensated own substitution effect suggests that higher wages, by increasing the opportunity cost of leisure time, will induce additional work effort. Also since religious activity tends to be more time intensive than other consumption commodities, higher wage rates may induce individuals to substitute other consumption for religious participation. Empirically, Azzi and Ehrenberg find significant negative wage effects on church membership rates: the substitution effect dominates.
From a time series perspective, secular real wage growth would lead us to expec that, as the opportunity cost of time rises, individuals substitute away from church attendance. Here, then, real consumption wages are operating as the relative price of church membership variable.
Conversely, exogenous non-labour income has an unambiguous positive (pure income) effect on church membership, contingent on its normal good status. Higher levels of non-wage income reduce work effort, releasing time for other pursuits.
Cyclically, as unemployment rises, the opportunity cost of investing in religious capital declines for those individuals displaced into non-market activities. The theory would therefore predict a counter cyclical church attendance pattern.
Beyond the parameters of the household allocation of time model, economists hav offered several hypotheses of relevance, though there is nothing which could be labelled a systematic theory. In particular, Hull and Bold (1987) recognize tha the church competes in a product market with other churches, religions and secular organizations for the allegiance of consumers. The improvement of entertainment technology has fostered the expansion of the consumption set in the post-war era. This entry by producers of activities which may substitute fo church involvement is argued to have precipitated secularization. Gilbert (1980 pp. 96-7), for example, posits that the emergence of mass television and motor vehicle ownership has undermined organized religion by transforming patterns of leisure activity. According to Gilbert, the television has functioned to privatize leisure, shifting recreation away from the public sphere, including public acts of worship, into the home. In tandem, the greater private mobility provided by the automobile has facilitated the social repositioning of leisure from the residential community to that of the family.
Hull and Bold also point out that, historically, churches have provided an important class of public and social goods: property rights enforcement, education; poverty relief and health care. However, organized religion is relatively less effective in the provision of such goods in industrial states due to the role of central government in the redistribution of income and the supply of health and educational services. The church is thus predicted to be more influential in complex societies without a fully developed central government.
The crowding out effect of government expenditure on church membership is teste by Wuthnow and Nass (1988) across US states. Their results indicate that expanding the functions of the welfare state does erode religious participation The main mechanism proposed to explain the negative association is the simple transfer of social functions, such as education and welfare, from the churches to the state. In the Scottish case, both Highet (1950) and Brown (1987) concur that the major social role of the Church of Scotland in education and poor relief elapsed during the inter-war period. This period marked the professionalization of such social services under the auspices of the civil state. Whether the argument can be extended into the post-war era is a priori unclear. It is tested below.
At least one British study (Wadsworth and Freeman, 1983) finds that higher education has played a powerful role in religious defection. This is significan given our earlier result that it is the decay in the rate of youth inflow into church membership which is the chief source of decline in the Church of Scotland. During the last 40 years there has, of course, been a massive expansion in higher education. Wadsworth and Freeman attribute its secularizing effect to disengagement from the family|3~ and an intellectual environment unsympathetic to religion.
The rather disparate theoretical hypotheses put forward to account for secular shifts in ecclesiastical disaffiliation motivate the variables' included in the following specification: INFLOW = f(WAGE, CYCLE, BILLY, CAR, GOVT, UNIV).
* INFLOW = Church membership inflow rate
* WAGE = Average real wages
* CYCLE = Business cycle variable
* BILLY = Dummy variable for Billy Graham mission in Scotland
* CAR = Number of private car licences per capita
* GOVT = Real public expenditure per capita
* UNIV = Proportion of young people studying at university.
The full definition and sources of these variables are reported in the Appendix The left hand side variable is specified in terms of the inflow rate due to the central role of inflows in explaining the variation in the church membership stock. WAGE and CYCLE are the key economic factors suggested by the allocation of time model. The wage variable relates to adult males. This is a relatively crude proxy for the ideal measure, a sex-weighted, youth wage variable. The business cycle variable consists of the residuals from the regression of the lo of Scottish real national income in a quadratic time trend. Note that a non-wag income variable is not included among the regressors. This is due to the absenc of appropriate Scottish data which could serve as a proxy. BILLY represents a dummy variable set to one in 1955 and 1956 to pick up the effects on membership inflows of the Billy Graham mission to Scotland in 1955. CAR attempts to model growth in consumption opportunities; GOVT picks up the crowding out of church functions by government activity; and UNIV is included to detect the proposed impact of higher education on church inflow rates.
The expected signs on the coefficients of CYCLE, CAR, GOVT, and UNIV are negative, that of BILLY is positive, and the sign on the parameter of WAGE is indeterminate a priori.
The equation was estimated in log-linear form by OLS using annual data for the period 1952-1988. Thus the coefficients represent estimates of the elasticity o the inflow rate with respect to each regressor. The estimation period was determined by data availability considerations. To introduce some dynamics into the model, each of the continuous independent variables was also specified with a one period lag. This general equation was then simplified by restricting to zero the coefficients of variables with t-ratios less than unity in absolute value.
The estimates and diagnostic statistics are reported in the first column of Table II. The overall performance of the equation is good. Using an F-test, the simplifying linear restrictions cannot be rejected: F(5, 24) = 1.125 compared with a 5 per cent critical value of 2.62. No significant autocorrelation can be detected and the equation easily passes a Chow predictive failure test over a split at 1970 (the 5 per cent critical value is 2.68).
The estimated coefficients are of the expected signs and their magnitudes are reasonable. Most variables are significant at the 5 per cent level, except for |GOVT.sub.t-1~ which is significant at the 10 per cent level and |UNIV.sub.t-1~ which is not statistically significant. The car ownership variable, CAR, was poorly determined and dropped early in the simplification process. Recognizing that the regression was based on just 36 observations, these would appear to be very satisfactory results.
Table II. Inflow Function Estimates, 1953-1988 log |INFLOW.sub.t~ log |(INFLOW/1-INFLOW).sub.t~ Constant 5.72 (6.55)(*) 6.11 (6.58)(*) |WAGE.sub.t~ -0.53 (1.91)(*) -0.53 (1.80)(*) |WAGE.sub.t-1~ -0.81 (2.87)(*) -0.83 (2.79)(*) |CYCLE.sub.t~ -0.55 (3.26)(*) -0.54 (3.00)(*) |BILLY.sub.t~ 0.16 (5.21)(*) 0.18 (5.61)(*) |GOVT.sub.t-1~ -0.16 (1.64) -0.16 (1.61) |UNIV.sub.t-1~ -0.11 (1.17) -0.19 (1.83)(*) |R.sup.2~ adjusted 0.995 0.995 DW 1.85 1.86 SER 0.0341 0.0362 CHOW 0.847 0.7306 F(18, 11) Note: DW is the Durbin Watson statistic; SER is the standard error of the regression; CHOW F(18, 11) represents a predictive failure test based on a sample split in 1970; * denotes significance at the 5 per cent level or better, using a one-tailed test.
Given that predicted values of |INFLOW.sub.t~, a proportion, should be restricted to lie in the (0, 1) interval, results are also reported for a logit transformation of the dependent variable. These are listed in the second column of Table II. They are almost identical to the log-linear estimates except for the coefficient on |UNIV.sub.t-1~ which is now significant at the 5 per cent level using a one-tailed test.
The results are consistent with our theoretical priors. The long run elasticity of the inflow rate with respect to the real wage is high at about -1.35. This agrees with US studies in which the negative substitution effect of a rise in wages dominates the positive income effect on church attendance.
The failure of the car ownership variable was disappointing. One would expect the expansion of the consumption set in post-war period to have acted as a powerful distraction to church involvement. It is likely, of course, that car ownership and the emergence of alternative consumption opportunities are strongly collinear with income. Its effect may therefore be difficult to separate from that of the real wage variable.
The prediction of countercyclical variation in church membership rates cannot b rejected by the data. There also appears to be some substance to the notions that the expansion of both the welfare state and higher education have had a negative impact on flows into membership of the Church of Scotland, though thes effects are not powerful.
There is no well-established theoretical or empirical literature explaining the dramatic post-war slide of church membership in the UK or, more specifically, i Scotland. Wolfe and Pickford (1980) studied the Church of Scotland from an economic perspective, but did not succeed in their attempt to construct an econometric model of membership. In contrast, we find that, in a simple OLS regression equation, most of our variables are correctly signed and significant
The unabated decline in the stock of Church of Scotland members since 1957 primarily reflects falling inflow rates among young people. Economic and socio-economic factors can be adduced to explain this observation. The secular rise in real wages has prompted a substitution away from church membership, though the precise mechanism through which the effect operates is difficult to specify. It could, for example, reflect the substitution of work for leisure, especially among married women re-entrants to the labour force, or the post-war expansion of the consumption set. We also find some weak support for the hypotheses that the growth in both higher education and the welfare state have had a negative impact on church attendance.
On the positive side, it would seem that both mission and recession stimulate increasing inflows into membership: the visit of Billy Graham in 1955 had a statistically significant effect, and inflow rates were found to vary countercyclically. However, it remains the case that the secular trend is downwards and, if real wage levels continue to play the key role, is apparently irreversible.
1. An important sample survey of Scottish church membership and attendance was also undertaken by John Highet in 1959 (Highet, 1960).
2. Transfer from the adult Catholic population is assumed, plausibly, to be negligible. The inclusion of both non-Church of Scotland Protestant church members and attenders in the non-member deflator is slightly less satisfactory. What constitutes potential church membership is inevitably ill-defined at the boundaries.
3. Presumably less important in the case of the Glasgow universities given the large number of students who live at home.
Anderson, G.M. (1988), "Mr Smith and the Preachers: The Economics of Religion i the Wealth of Nations", Journal of Political Economy, Vol. 96 No. 5, pp. 1066-88.
Azzi, C. and Ehrenberg, R. (1975), "Household Allocation of Time and Church Attendance", Journal of Political Economy, Vol. 83 No. 1, pp. 27-56.
Brierley, P. and MacDonald, F. (1985), Prospects for Scotland: Report of the 1984 Census of the Churches, Marc Europe and the National Bible Society of Scotland.
Brown, C.G. (1987), The Social History of Religion in Scotland since 1730, Methuen, London/New York.
Ehrenberg, R. (1977), "Household Allocation of Time and Religiosity: Replicatio and Extension", Journal of Political Economy, Vol. 85 No. 2, pp. 415-23.
Gilbert, A.D. (1980), The Making of Post-Christian Britain, Longman, London.
Highet, J. (1950), The Churches in Scotland Today: A Survey of their Principles Strength, Work and Statements, Jackson Son & Company, Glasgow.
Highet, J. (1960), The Scottish Churches: A Review of their State 400 Years after the Reformation, Sheffington, London.
Highet, J. (1985), "Trends in Attendance and Membership", in Brierley, P. and MacDonald, F. (Eds) Prospects for Scotland: Report of the 1984 Census of the Churches, Marc Europe and the National Bible Society of Scotland.
Hull, B.B. and Bold, F. (1989), "Towards an Economic Theory of the Church", International Journal of Social Economics, Vol. 16 No. 7, pp. 5-15.
Iannaccone, L.R. (1990), "Religious Practice: A Human Capital Approach", Journa for the Scientific Study of Religion, Vol. 29 No. 3, pp. 297-314.
Long, S.H. and Settle, R.F. (1977), "Household Allocation of Time and Church Attendance: Some Additional Evidence", Journal of Political Economy, Vol. 85 No 2, pp. 409-13.
McCrone, G. (1965), Scotland's Economic Progress 1951-1960, George Allen & Unwin, London.
Parliamentary Papers (1854), Religious Worship and Education: Scotland, Report and Tables, LIX.
Ulbrich, H. and Wallace, N. (1984), "Women's Work Force Status and Church Attendance", Journal for the Scientific Study of Religion, Vol. 23 No. 4, pp. 341-50.
Wadsworth, M.E.J. and Freeman, S.R. (1983), "Generation Differences in Beliefs: A Cohort Study of Stability and Change in Religious Beliefs", The British Journal of Sociology, Vol. 34 No. 3, pp. 416-37.
Wolfe, J.N. and Pickford, N. (1980), The Church of Scotland: An Economic Survey Geoffrey Chapman, London.
Wuthnow, R. and Nass, C. (1988), "Government Activity and Civil Privatism: Evidence from Voluntary Church Membership", Journal for the Scientific Study of Religion, Vol. 27 No. 2, pp. 157-74.
INFLOW Inflow into church membership as a proportion of the population aged 17-21 (CYSB, RGS).
WAGE Adult manual male hourly earnings in Scotland deflated by the retail price index (SAS, SEB, AAS, ETAS).
CYCLE Residuals from the regression of the log of Scottish real national income on a quadratic time trend (SAS, McCrone, 1965).
BILLY Dummy variable for the Billy Graham mission, set to one for 1955 and 1956
CAR Current car licences per capita in Scotland (SAS, RGS).
GOVT Public expenditure per capita in Scotland deflated by the GDP deflator (SAS, RGS).
UNIV Students in Scottish university education as a proportion of the populatio aged 17 to 21 (SAS, RGS).
AAS = Annual Abstract of Statistics CYSB = Church of Scotland Year Book ETAS = Economic Trends Annual Supplement RGS = Registrar General Scotland, Annual Report SAS = Scottish Abstract of Statistics SEB = Scottish Economic Bulletin
|Printer friendly Cite/link Email Feedback|
|Publication:||International Journal of Social Economics|
|Date:||Dec 1, 1993|
|Previous Article:||On need, wants, resources and limits.|
|Next Article:||Sir Richard Arkwright: premier crook in the Industrial Revolution.|