1 VATT-KESKUSTELUALOITTEITA VATT-DISCUSSION PAPERS 351 FINNISH HOUSEHOLD CONSUMPTION IN MONETARY AND PHYSICAL TERMS TRENDS AND CLARIFICATIONS Adriaan Perrels Risto Sullström Valtion taloudellinen tutkimuskeskus Government Institute for Economic Research Helsinki 2004
2 ISBN (nid.) ISBN (PDF) ISSN (nid.) ISSN (PDF) Valtion taloudellinen tutkimuskeskus Government Institute for Economic Research Arkadiankatu 7, Helsinki, Finland Oy Nord Print Ab Helsinki, November 2004
3 PERRELS ADRIAAN SULLSTRÖM RISTO: FINNISH HOUSEHOLD CONSUMPTION IN MONETARY AND PHYSICAL TERMS TRENDS AND CLARIFICATIONS. Helsinki, VATT, Valtion taloudellinen tutkimuskeskus, Government Institute for Economic Research, 2004, (C, ISSN (nid.), ISSN (PDF), No 351). ISBN (nid.), ISBN (PDF) Abstract: The present report provides an overview of the development of household consumption in Finland in the 20 th century. The study has been carried out in the framework of the AESOPUS study funded by the Academy of Finland. The AESOPUS study aimed to map out the stocks and flows of nitrogen and phosphorus in Finland, as well as to analyse the man made influences on those flows. For this reason special attention was paid to the consumption of foodstuffs in the VATT study. Next to an overview of the developments during the entire 20 th century, various features with respect to the last decades are discussed, based on micro-data available for years in the period Apart from describing the developments in consumption, the report also highlights the evolution of main driving forces behind the consumption developments. Actual estimation of consumption functions and the construction of a simulation model are treated in a separate report. Key words: consumption, households, food PERRELS ADRIAAN SULLSTRÖM RISTO: FINNISH HOUSEHOLD CONSUMPTION IN MONETARY AND PHYSICAL TERMS TRENDS AND CLARIFICATIONS. Helsinki, VATT, Valtion taloudellinen tutkimuskeskus, Government Institute for Economic Research, 2004, (C, ISSN (nid.), ISSN (PDF), No 351). ISBN (nid.), ISBN (PDF) Tiivistelmä: Tämä raportti luo yleiskuvauksen kotitalouksien kulutuksen kehityksestä Suomessa 1900-luvulla. Tutkimus toteutettiin osana Suomen Akatemian rahoittamaa AESOPUS -hanketta. Hankkeessa kartoitettiin typen ja fosforin virtoja ja varantoja Suomessa. Lisäksi tutkimushankkeessa analysoitiin ihmisen toiminnan vaikutuksia virtoihin. Tästä syystä VATTin tutkimustyössä keskityttiin erityisesti elintarvikkeiden käyttöön. Yleiskuvauksen lisäksi raportissa tarkastellaan mikroaineiston avulla muutamaa yksityiskohtaa vuosina Raportissa huomioidaan myös kulutuksen kehityksen taustavoimat. Varsinaiset kulutusyhtälöiden estimoinnit ja kulutus/panostuotos -malliyhdistelmän simuloinnit käsitellään erikseen toisessa raportissa. Asiasanat: elintarvikkeet, kotitaloudet, kulutus
4 Summary Context Consumption of households in Finland has been studied in the framework of a project concerning the description and analysis of flows and stocks of nitrogen and phosphorus in Finnish nature. The project, AESOPUS (Analysis of nutrient cycles in Ecological and SOcio-economic systems for Policy PUrposeS), aimed to acquire a complete picture of the current stocks and flows of phosphorus and nitrogen, as well as obtain an impression of its historical development in the past 50 to 100 years. In addition the project investigated the main sources of man made intervention in the flows and stocks and how these interventions may be altered for the purpose of reducing the loads in fresh water systems and eventually of limiting or even extinguishing eutrophication in the Baltic Sea. AESOPUS was one of several studies of which the Finnish Academy programme SUNARE was composed of. AESOPUS was co-ordinated by the Finnish Environmental Institute (SYKE). Other partners were the Agricultural Research Centre of Finland (MTT), the Finnish Forest Research Institute (METLA), Helsinki University, and the Government Institute for Economic Research (VATT). Next to analysis of overall economically induced impacts on emission levels by means of an extended input-output model, VATT s contribution concerned the analyses of consumption patterns, notably of food. However, also other items that cause directly or indirectly significant emissions of nitrogen and/or phosphorus (such as private car ownership and use) are considered. The present report provides an overview of the developments in consumption in the past century as well as an impression of the main driving forces behind the developments. The estimation of a consumption expenditure mode and the construction of and simulations with an input-output model are discussed in a separate publication. The present report is supposed to be accessible for a larger audience. Therefore the authors tried to minimise on technicalities. The other report will have a more technical-nature. Findings The construction of longer uninterrupted time series of more detailed observations of consumption appears not easy to realise. This is even more so if next to expression in monetary units also physical information (volumes) is sought for. Several studies are available, each covering a good part of the information but none of them are complete. Furthermore, at more detailed levels and in case of physical data, the various sources may come up with different numbers for the same year for the same item. The last 25 years the reliability has improved, for example, divergence between consumption and production based estimates of food intake seem to diminish substantially.
5 Also the since 1966 recurrently performed consumer expenditure surveys among several thousands of households have created a valuable basis for more detailed longitudinal analysis, especially with respect to the better comparable surveys of 1985, 1990, , 1998 and It is regrettable and from the point of view of sustainability research and policy preparation downright deplorable, that since 2001 the consumer expenditure surveys do not anymore include systematic observations of physical amounts purchased. Wealth measured in monetary terms has increased enormously. In real terms (i.e. corrected for inflation) annual consumer expenditures per capita increased almost tenfold between the beginning and the end of the last century. This process went hand in hand with large changes in the settlement patterns. From a population predominantly living in rural settlements the situation transformed into a population predominantly living in medium-sized and larger cities. As regards foodstuffs this meant that whereas in 1900 a significant part of the population had access to non-market products (e.g. from own plots, from gifts or obtained through barter) most households in 1998 were to an overwhelming extent relying on purchased foodstuffs. This transformation accelerated after Per capita expenditures on food in real terms increased by approximately a factor 2, which is much less than the factor 10 for the overall budget. The largest percentage increases of the food budget occurred before The above mentioned conversion from non-market to market provision of foodstuffs will probably explain a part of the slow down after Another expenditure category relevant for nitrogen emissions is transport. This category was still very modest in 1950, with a budget share of 7 %. In 1990 this share had gone up to 17 %, whereas transport expenditures per capita had increased by a factor 9 times in real terms since For comparison, the budget share of foodstuffs was 29 % in 1950 and went down to about 12 % in After 1990 the expenditures on transport grow at a much slower pace. Even though changes in food consumption look less dramatic at the macroeconomic level compared to some other categories, a closer look at developments within the category food shows still considerable changes. For example, more has been spent on meat, fruit and vegetables. Meat embodies much higher environmental impacts than vegetables. Comparison of physical and monetary data also shows that trends can look really different when considered in physical terms instead of monetary ones, which is particularly relevant in relation to environmental applications. With respect to meat it seems that increased expenditures per capita largely coincide with more kilos per capita. However, in the case of dairy products increased expenditures per capita seem to relate to smaller overall amounts and a shift from basic to more processed dairy products. Similarly for vegetables the total amounts (potatoes included) remains by and large the same,
6 but to some extent potatoes are substituted for other vegetables, of which an increasing share is imported. After 1985, and notably after 1990, net disposable income per household became less evenly distributed. The changed structure that starts to emerge after 1995 implies that employed middle aged persons received most of the improvements of purchasing power. Even though the crisis of the early nineties also hit elderly employees in particular a cohort of unemployed or weakly employed people currently in the ages (of household heads) travels through the purchasing power distribution over time. Up to an age of about 32 average annual expenditures exceed average annual disposable income. Despite the remarkable changes in distribution patterns, the age at which average disposable household income starts to exceed average annual household expenditures has barely changed. When comparing the disposable incomes and household expenditures per capita by age category it seems that disposable income dynamics and expenditure dynamics have not been following entirely the same dynamics. In 1990 disposable household income per capita was on average the highest for age class of 55. Since then this highest point moved to the age of 58. Interestingly for the average expenditures per age category the maximum is at the age of 51 in 1990, whereas this maximum moves to 57 in years following In general the distribution of average expenditures by age category is less uneven than is the case for disposable income. When applying Engel curves in which budget shares per expenditure category are related to age of adult household members it appears that different categories show different levels of stability of the Engel curves. Categories containing higher shares of so-called luxury goods (like transportation ) tend to show more significant changes in the slope of the Engel curves, thereby pointing at essential changes in propensities to consume and the underlying preference structures. Even though such changes may look modest, they have major ramifications for the long term developments of household consumption.
7 Contents 1. Introduction 1 2. Description of data sources and types Consumption surveys of household expenditures Household consumption in physical amounts Aggregate consumption series Price data 9 3. Generic developments in household consumption The various meanings of consumption Historical review of some important background factors Population size and age structure Urbanisation Purchasing power and wealth Developments in consumption patterns Long term trends in expenditures Long term trends in physical consumption Explanatory factors in recent decades Consuming versus saving Propensity to consume Engel curves Synthesis trends, factors, implications for the estimations 33 References 36 Appendix 1 AESOPUS project proposal 38 Appendix 2 Classification of expenditure categories in household budget 49 surveys Appendix 3 Concentration curves and Engel curves 64
9 1. Introduction This publication discusses the findings of the first phase of VATT s share in the socalled AESOPUS project. Prior to clarifying the purpose of especially VATT s contribution, it is proper to briefly introduce the entire context of the study, which in turn helps to clarify choices of focal areas and of methodologies in the analysis of consumption. The AESOPUS project aims to produce a comprehensive systematic analysis of the nitrogen (N) and phosphorus (P) cycles in Finland at national and regional levels. Furthermore, the project aims to analyse the impact of consumer and industrial behaviour and the impacts and options of policy interventions on the N and P cycles. Human activities have greatly increased the fluxes of major nutrients (carbon, nitrogen and phosphorus) in the natural environment. The amount of nitrogen in the biogeochemical cycle has doubled through nitrogen fixation in the fertilizer industry, burning of fossil fuels and increased use of leguminous crops in agriculture (Jordan and Welder, 1996; Vitousek et al., 1997). As a consequence of those processes and applications the fluxes of nitrogen and phosphorus from land to oceans has greatly increased (Carpenter et al., 1998). The most severe implications of these disturbances are climate change (mostly related to the carbon cycle), eutrophication and acidification of aquatic ecosystems, and pollution of groundwater, imbalances in soil nutrient status, depletion of non-renewable resources and non-sustainable use of renewable resources. The nutrient cycles consist of renewable and non-renewable stocks, and fluxes between these stocks. Human disturbances of the natural cycles are driven by production (agriculture, forestry, transport, energy and other industries) and consumption sectors. The utilisation of the stocks defines the magnitude of the fluxes, but may also have an impact on the stocks. The magnitude of human-induced fluxes and changes in stocks is affected by market forces, policy interventions, technology choices and consumer preferences, while perceptions, values and beliefs play a steering role in the background. Consumer behaviour is a driving force for production sector and policy focus; and ultimately also in nutrient cycles. Thus information on the present and historical magnitude of fluxes and changes in stocks as well as understanding the impact of consumer behaviour and policy interventions are important for the more sustainable use of natural resources and for better environmental quality. As the overall project aims at comprehensive descriptions of stocks and flows for N and P, it also necessary to aim for a comprehensive description of consumption and its impacts on product and material use during the production stages in industry, agriculture and mining. Furthermore, as the overall project also intends to assess impacts of
10 2 policy interventions, it is required to device some formalised description of how and how much households consume and what main factors are influencing the consumption patterns (Appendix A.1 contains the research plan). Following from these requirements it is clear that some kind of modelling system for household consumption both in terms of expenditures and acquired volumes is necessary. To this end a set of consumption functions has to be estimated. The consumption functions can subsequently be used to simulate consumer behaviour and to feed back the results into a macro-economic model system, being an input-output model or otherwise. The present report concerns the description of trends and explanatory factors of Finnish household consumption in monetary and physical terms. The core of the data is formed by the various consumption expenditure surveys as they have been conducted by Statistics Finland since In addition other data have been used from Statistics Finland, the Thule Institute and VATT. Special emphasis is put on those consumption items that are most relevant to impacts on N and P flows. In practice this means food and drinks and energy use (including transportation fuels). The current report will not present a newly estimated set of consumption functions, as that will be dealt with in a separate report. However, some preparatory work related to the estimations will be discussed. This report has been written with the intention to keep it accessible also for those not familiar with economics. The follow up report on estimations of consumer expenditure system and the use of these estimations in a simulation model, will have of necessity a more technical nature.
11 3 2. Description of data sources and types Though the econometric analysis of micro-data sets are the main purpose of the present study, there has been added a concise retrospective survey of consumption developments since the beginning of the 20 th century. For the retrospective survey data have been used from Laurila (1985) for the period as well data from the National Accounts of Statistics Finland for the period A summary of data sources is given in table below. An important difference is that the sources mentioned under (1) and (2) in the table below are micro-data, allowing the researcher more degrees of freedom and possibilities to disentangle impacts better. The sources mentioned under (3), (4), (6) and (7) concern aggregate and reprocessed data, which include assumptions and corrections that make it more difficult to disentangle impacts and attribute them to influence factors. This applies especially to the last two sources. Table 2.1 A summary of data sources Data source Sample-based Consumption Expenditures Surveys (CES) (1) * (2) (3) Sample-based consumption surveys of purchased volumes (same source as 1) * Aggregate expenditure time series of the System of National Accounts (SNA) at current and constant prices * Retrospective survey Econometric analysis (4) Final consumption of I/O-tables * X (5) (6) (7) Monthly price data by commodities * Synthesised long term time series of expenditures, prices, selected physical volumes ** Additional time series on physical consumption (food, validation purposes) *** Source: *) Statistics Finland; **) Laurila (1985); ***) Heikkinen and Maula (1996); Leppälä (1992) X X (X) X X X X
12 4 In order to estimate consumption behaviour and connecting it into the model of AE- SOPUS we need first suitable data and secondly ways to link data and results from different sources together. 2.1 Consumption surveys of household expenditures The main source of data for the study consists of household budget data collected through Consumption Expenditure Surveys (CES) conducted by Statistics Finland. The first budget survey based on representative random sampling was carried in After that extensive surveys have been collected at regular about five years intervals in 1971, 1976, 1981, 1985, 1990, , 1998 and The survey of consists of the cross-sections of three years, 1994, 1995 and The most useful surveys for the intended econometric analysis in this study are the latest ones, i.e. the surveys of 1985, 1990, , 1998 and The classification of variables in the data sets of is in accordance with the COICOP-classification (Classification of Individual Consumption by Purpose). Nevertheless the originally data sets for the years 1985, 1990, , 1998 and 2001 had some minor differences in variable definition, as well as slight differences in the numbers of background variables and consumption sub-categories. As much as possible these differences have been harmonised, whereas the price-index data sets have been harmonised as much as possible with consumption expenditure survey categorisation. By doing so it is possible to analyse the cross-section data-sets of different years in conjunction, and thereby try to distinguish between cross-sectional 1 and longitudinal effects. The consumer expenditure surveys of Statistics Finland constitute the principal source of information on overall and detailed expenditure behaviour of individual Finnish households in connection with a very wide selection of household characteristics. The data are collected using personal interviews, household consumption accounts and administrative registers. The data sets contain approximately 170 background variables for household characteristics (household size, age, education level, employment status, dwelling type, regional location, etc.), many income items and distinguish about 800 expenditure (sub)categories for which annual expenditure by household (in the survey) is recorded (see Appendix 2). The net sample sizes, total non-responses and the length of bookkeeping periods are given in Table 2.2. In 1981, the size of sample was approximately 7400 households, in 1 Cross-sectional effects refer to influences of variables when comparing different household types in the same observation year (e.g. household size and age). Longitudinal effects refer to the impact of changes in circumstances for the various household types over time, the prime influence is that of the evolution of household disposable income and relative prices of products, but in the background also other developments count, such as urbanisation and rising education levels (both are background variables). Yet, changes in norms and values, fashion influences, etc. have impact as well, but can not be singled out.
13 , in , in about 6700 and in households. Other surveys are clearly smaller. Since the year 1981 the bookkeeping periods consists of two weeks. We see that at its introduction a shorter bookkeeping period clearly cuts down non-respond times, later on the effect fades away. The expenditure figures recorded in the data sets however represent annual amounts, obtained by applying an augmentation procedure, which among others accounts for product specific features and seasonal impacts. Table 2.2 Consumption Expenditure Surveys in Finland Year Net sample size Non-response % Bookkeeping period one month one month one month two weeks two weeks two weeks two weeks two weeks two weeks two weeks two weeks Source: Consumption Expenditure Surveys, Statistics Finland 2.2 Household consumption in physical amounts The consecutive consumption surveys of Statistics Finland did not only produce a data set on expenditures, but also a parallel data set on the respective purchases expressed in kilos, litres, or pieces 2. Those volume data sets have basically the same structure and categorisation as the expenditure files have. Nevertheless in practice some harmonisation of commodity categories appeared necessary. Furthermore, in order to combine kilo, litre and piece items various transformations are needed. We use only the volume data sets of the years 1985, 1990 and In order to make a connection between economy and ecology the availability of parallel data on expenditures and physical volumes is essential to be able to translate economic behaviour of households more reliably into physical impacts. Increased expenditures for a particular category, corrected for inflation, do not necessarily mean that consumption in physical terms has increased, even though the default economic description would be that consumption of commodity A has increased in real terms. 2 Many foodstuffs are expressed in either kilos or litres, and sometimes as pieces. It required additional work however to obtain a transparent set of volumes used for all purchases. 3 Airi Pajunen from Statistics Finland gave important advice in transforming and harmonising the data. Without this knowledge reliable comparison would have been impossible.
14 6 Even more complicated is the possibility that the quality of commodities consumed can change even if neither physical amounts nor inflation corrected expenditures did change. Since prices-index data sets are available as well it is to some extent possible to separate quality changes from volume and price changes. The application of this technique (Deaton, 1987; Brubakk, 1997) will be discussed in the next report on estimations. Especially with regard to the intake of food and beverages it is hard to produce a consistent and complete overview. Households do not only purchase foodstuffs which are eaten at home, but they also enjoy meals in restaurants and in cantinas in their work place. In contrast to food purchases in shops, the categorisation of restaurant meals and the like is much less refined, and also includes a sizeable category mixed/not attributable. Furthermore, households may receive servings as part of a larger service package (i.e. during holiday events), whereas they receive them for free during free time and work related visits. Neither of these options is (nor can be) observed in the household consumption surveys. An additional source of deviations is that a part of the households produces some foodstuffs from own horticulture and/or from picking in forests. In principle this option is observed in the household consumption surveys, but not very accurately. Some of the acquisitions are termed as free consumption and others as income in kind. An estimated unit value has to be attributed to these self produced items. In some cases market prices will be adequate approximations, in other cases not. The approximate average share of those items in the total consumption budget is about 0.6 %. For foodstuffs the share is over 2 % of the food expenditures, with large variations between urban and non-urban residents. For inhabitants in larger cities the share hovers around 1 %, whereas for people living in the countryside the average share approaches 5 %. For sub-categories within the food budget, such as fruit and vegetables, the shares of free consumption are probably even substantially higher. Last but not least there is usually underreporting of some purchases, this affects especially two sub-categories (see also next section): - small items bought separately, such as candy bars - alcoholic beverages. Considering the various uncertainties and biases mentioned, it can be concluded that the consumption survey is in the first place an expenditure survey, notwithstanding the availability of physical volumes obtained, For analysis of the physical volumes consumed, auxiliary information is necessary. It also means that later on in the econometric analysis not all substitution alternatives can be mapped appropriately, despite the possibility to distinguish quality effects to some extent.
15 7 2.3 Aggregate consumption series As already indicated in section 2.1 and 2.2 additional micro and notably macro level information is required to check differences between the aggregate expenditures of households based on the Consumption Expenditure Survey (CES) and aggregate consumption as reported in the national accounts (NA). Comparisons between the NA and estimates based on aggregation of the CES data are given in Table 2.3. We can see that consumption expenditures according to the CES are lower than reported in the NA. The range of the deviation between the assessments of total expenditure lies between 88.6 % (1985) and 92.5 % (1990). In addition to under-recording of some categories in the survey, the samples of the CES do not include persons in institutions, whereas NA does. An additional reason relevant at category and sub-category level is the inadequacy of the augmentation procedure to capture all distributional features when simulating annual figures per household (more discussion below). Table 2.3 Individual consumption expenditures by category, compared for NA and CES (expressed as: 100*CES/NA) Main commodity groups % inst. Total expenditures of which for: Food Beverages and tobacco Clothing and footwear Housing and energy Furnishings and household maintenance Health Transport Communications Recreation and culture Education Hotels, cafes and restaurants Miscellaneous goods and services Source: Consumption Expenditure Surveys (CES), National Accounts (NA), Statistics Finland. The last column of table 2.3 shows the significance of the population in institutions in various categories. The table also shows that aggregation from CES may also lead to overestimation of national figures, as is for example the case for housing in 1990 and We see that in the case of food that the ratio of the two results varies from 90.8 in 1985 to 99.2 percent points in 1995 and Especially severe underestimates are indicated for the categories beverages and tobacco, education and hotels, cafes and restaurants. Remarkable is also the significant deterioration of the fit after 1990 for the category miscellaneous ( beauty services, jewellery and accessories, cleaning services, social services and counselling and financial services ).
16 8 The augmentation procedure for each commodity (sub)category to produce annual expenditures can only be applied if households report a purchase in the respective (sub)category during the survey period (two weeks). In other words if there is a zeroobservation, no annual figure can be given for that household. The larger and more expensive an item is and the more durable it becomes, the higher is the likelihood that a household does not purchase that commodity or service during the two-weeks survey period. If the survey is large enough and if purchase habits (including frequency) did not change, the zero-observations are not a serious obstacle for assessing annual expenditures by type of household (provided each household type has sufficient representation in the survey). However, in case of changes in purchasing behaviour, notably frequency and seasonal influences, the augmentation procedure may result in biased annual expenditures. Similarly, if the fraction of the survey population with zeroobservations becomes very high it can also lead to biased aggregates. Table 2.4 summarises the occurrence of zero observations by expenditure category in subsequent survey years. The fractions of zero-observations in the samples of are high for the categories beverages and tobacco, clothing, health, education, and hotels, cafes and restaurants. The high fraction for education is due to the very dominant public provision of education, making education as commercial service for households a fairly marginal affair. Moreover a part of the private education will be recorded under culture. The 2001 survey shows declining fractions of zeroobservations. Especially for the categories clothing and footwear and transport the reduction is substantial. This is probably a reflection of continued improvement of purchasing power combined with restored consumer confidence. Also the downside of the boom and bust development in nineties shows especially in the zero-observations for clothing and footwear. The first jump between 1990 and 1995 is party attributable to reduced purchasing power resulting from the sharply rising unemployment in the period The continued increase of the fraction between 1995 and 1998 (when employment and purchasing power had recovered to some extent) hints however also at changes in purchasing behaviour, meaning that probably both the frequency in clothing purchases has decreased and clothing has gone down in the preference ranking of households.
17 9 Table 2.4 Zero observations as share of households in which purchases in that category were not reported (%) in CES Commodity groups Food Beverages and tobacco Clothing and footwear Housing and energy Furnishings and household maintenance Health Transport Communications Recreation and culture Education Hotels, cafes and restaurants Miscellaneous goods and services Source: CES micro data sets 2.4 Price data A deficiency in CES is lack of price information. The household survey data only give annual expenditure per item, but no unit price. Therefore a separate file has been obtained containing the time series of the Consumer Price Index for a large number of commodities covering the period We created a correspondence between the two-week bookkeeping periods and monthly price indices by means of linear interpolation in those surveys where the two weeks period were in effect, being 1985, 1990, 1995, 1998 and Furthermore, the commodity classification in the price index data files deviates from the one used in the household survey, especially at very disaggregate levels, affecting the foodstuffs in particular. The classification has been rearranged in order to match with the household survey data files, this meant sometimes loss of information due to aggregation. By combining the volume data and expenditure data into one file the apparent average unit price that a household has been paying for types of food, can be derived. Obviously these apparent unit prices, which are household specific, cannot be used (at least not straightaway) as alternative price series data. The relation and distinction between these unit prices and the observed prices as summarised in the price index plays an important role in the analysis.
18 10 3. Generic developments in household consumption 3.1 The various meanings of consumption Households combine goods, time and skills to produce a ready-to-consume product. Household have a range of choices whether to purchase commodities that need little own input (of either time and/or skills) or conversely purchase commodities (semifinished or even raw materials) that need a lot of time and skill input to end up with a ready-to-consume product. For example one can on the one hand buy prepared sandwiches and on the other hand purchase flour and yeast and make the bread one selves and combine that with products from the own garden (tomatoes, chicken eggs, etc.). The probably most common solution is to purchase bread, butter and tomatoes, etc from a shop and do only the final preparations oneself. The above description follows the view that in households a set of functions has to be performed in order to enable household members to actually derive utility from the endowments at their disposal. This concept is referred to as new home economics (see e.g. Becker, 1965; Winston, 1982; Gronau, 1986), which stresses that households (or its members) have often first to produce a ready-to-consume good or service before the household members can engage in the actual consumption itself. In addition a household can also take in ready-to-consume goods or services from commercial providers outside the household (e.g. movie theatres and restaurants). An alternative approach of describing the functioning of a household, which has closer links with the natural sciences, is the metabolism approach. In this approach the material input, transformation and output of households is described (e.g. Noorman and Schoot-Uyterkamp, 1998). In this study we will stick to the concept of consumption as meant in the (narrow) economic sense of purchasing commodities, albeit that we will extend the span of the analysis in order to capture physical volume developments as well. By doing so both the links with physical flows throughout the economy and the interaction with nature can be made more precisely and also offers possibilities for future extension of a full home economics model, encompassing money, physical resources, (labour)time and skills. On the other hand the chosen approach also ensures that linkages with macro-economic models are feasible, thereby allowing for testing for the overall economic impact assessment of nitrate and phosphorus policy measures operating on private consumption and the production incited by that consumption. Another issue, not further elaborated here, is the distinction between consumables, semi-durables, and durables. Consumables are used up within a relative short time span following on the purchase, and consequently consumables are purchased frequently (i.e. daily or weekly like for foodstuffs). Semi-durables are products used up over a somewhat longer time span (a month to a few years). Usually semi-durables are disposed off at the end of their technical lifetime, whereas repair is usually not possi-
19 11 ble or attractive (e.g. tires, rechargeable batteries, light bulbs, etc.). Durables can have long life times (few years to tenths of years), whereas repair is often attractive up to a certain lifetime of the product. Durables are purchased by the individual household on a very irregular basis, whereas the purchase is often a large expenditure (possibly partly or wholly paid out of non-regular income or savings). Given the focus on nitrate and phosphorus flows we are especially focussing on foodstuffs, and fuel use. Other items are of minor importance in this respect. The earlier mentioned choices between ways of provision are important for this study because: - They relate to urbanisation and the emergence of a consumer product industry, resulting also in growth of impact areas related to consumption-production chains. - Evolution in the choices of ways of provision also necessitates to be careful with interpretation of the results, as a seeming reduction in the use of a commodity may only mean it reappears, often more or less hidden (i.e. embedded or embodied) in another consumption category. - Important changes in the shares of the ways of provision can have also important policy implications, as these changes may require a reorientation across sectors or a renewal of the instruments used. The above considerations stress the importance of the interaction between the operational environment of a household and the emerging consumption pattern of a household. In addition to these physical contextual influences there are social-psychological and institutional influences on these emerging consumption patterns. Though at a very basic level people all over the world have similar needs (food, shelter, etc.), in practice the way these needs take shape can differ considerably due to differences in value systems between populations and within one population over time. Therefore, apart from relatively easy measurable changes such as in disposable income, household size, kind of dwelling and place of residence also the influence of advertisement, swings in fashionability of products and evolutions in opinions about kinds of products and habits affect consumption patterns. In this respect it is good to realise that even though this study is analysing consumption as an economic phenomenon, the other dimensions are just as relevant in terms of explaining the forces that affect the evolution of consumption patterns. For example, in modern sociology consumption is not so much any more about affluence or lack thereof, but more about a means to express oneself, i.e. underlining a typical identity and distinguishing from other typical identities (Bourdieu, 1979; Ritzer, 2001). The consequence of this is that in modern societies the changes in consumption patterns on the margin are significantly influenced by factors that are not or not sufficiently repre-
20 12 sented in the typical standard economic toolbox, even though consumption in its entirety is still bound by budget limits (and hence relative prices and income). Having discussed the various dimensions of household consumption as well as the background factors influencing these dimensions, the next section will briefly report on several important background trends such as development of the population and urbanisation. 3.2 Historical review of some important background factors Population size and age structure Demographics are an important factor in the evolution of consumption, both for aggregate private consumption and for the budget allocation. Age distribution and distribution over household types affect the composition of consumption. Figure 3.1 summarises the growth of the population from barely 2.7 million in the year 1900 up to almost 5.2 million at the end of the century. The population growth slowed down after 1960, but the number of households kept increasing faster, i.e. between 1960 and 1998 the population grew by approx , while the number of households grew by approx As a consequence average household size went down from 3.34 to 2.26 persons per household. The decrease in household size can be largely identified with increases in the number of one-parent households (often as consequence of divorce) and more single elderly people. Figure 3.1 Population size and its composition by age (1000) POP Source: Population statistics, Statistics Finland; VATT
21 Urbanisation An adjacent development has been the strong urbanisation, as is summarised in table 3.1. Municipalities with medium sized or small populations experienced stagnation or even reduction, unless they were situated in the neighbourhood of one of the larger cities (over inhabitants). The trend in the last part of the 20 th century has been a substantial net population inflow to the largest centres and a few successful medium sized municipalities. For a start it means that compared to pre 1960 decades ever more households could rely less on own produce. In addition it also means that ever more households became exposed to urban lifestyles, including the conscious and unconscious marketing of aspects of such lifestyles. As regards the food processing industry and the forest industry the patterns were not the same as for consumers. The concentration of production capacity in the paper industry did not imply a generic migration out of the medium sized and small mill towns, though some of these towns did loose industrial production. The food processing industry also went through a cycle of concentration of production (a process still continuing in the dairy industry). As proximity of households is often very relevant for the food industry, the (re)location pattern is following the geographical concentration of the population, albeit loosely (i.e. at quite aggregate spatial scales). Table 3.1 Change of population by type of municipality (1000) Municipal group % Change Pop Pop All municipalities 17 % The 100 largest municipalities in % All but capital area 16 % All municip. < inhabitants in % All municip. < inhabitants in % Capital area (Greater Helsinki) 80 % All municip. >30000 inhabitants in % All municip. > inhabitants in % Source: Population statistics, Statistics Finland; VATT Purchasing power and wealth Purchasing power of households has been growing tremendously over the past century. On a per capita basis real consumer expenditures grew more almost tenfold when comparing the 1900 level with the 2001 level (see figure 3.2) 4. At the aggregate level this amounts to about a nineteen fold increase due to the growth of the population. The 4 At this highly aggregate level an uninterrupted consistent time series was hard to construct, hence the interruption in In order to keep consistency with the next chapters concerning detailed consumption data from the same sources, the referred data are used. The harmonisation procedure implies that the development shown in figure 3.2 is a reasonable approximation of the exact figures.
22 14 improvement in purchasing power in the second part of the last century is about twice as large as the growth in the first part. The increase in purchasing power shows in all expenditure categories (see next chapter), but some categories showed much more growth in expenditures than other ones. The increasing expenditures have not only translated into more volume, but notably also in more variety and diversity. The improvement of purchasing power over time and the adoption of urban lifestyles also show in the next figure 3.3 in which ownership of durable consumer goods is shown for the period The figure includes especially items that relate to food habits and fuel use. Car ownership has been growing until the early nineties, and seemed to stabilise at about 65 % of the households. However, continued consumer confidence and recent changes in taxation of private cars, in particular of self-imported ones, has translated in a significant increase of car sales. Ownership rates should have increased beyond 70 % after 2001 (Ahlqvist and Berg, 2003; Mäkelä et al, 2002). In absolute terms the total car stock grows even slightly more because of the increase of the number of households. Considering the continuous increase of ownership rates of many appliances and the introduction of new applications, it is understandable that average electricity consumption per home keeps on growing and with decreasing family size also per capita. The spreading ownership of refrigerators, freezers, and microwave ovens reflect part of the changes in food habits as well as purchase habits (of food). Figure 3.2 Growth in the aggregate and the per capita real consumption budget Cons_hh Cons/cap Source: Laurila (1985) ( ); NA ( ), Statistics Finland.
23 15 Figure 3.3 Ownership of selected consumer durables, in % of households having at least one piece at their disposal color TV video player CD player PC refrigerator freezer microwave washing machine dishwasher internet connection car Source: Ahlqvist and Berg (2003).
24 16 4. Developments in consumption patterns 4.1 Long term trends in expenditures As was already demonstrated in the previous chapter increase in purchasing power has been impressive over the last century. Closer inspection of the growth in expenditures in real terms shows that on a per capita basis 5 expenditures on foodstuffs from shops have barely been growing since The same applies for beverages. Recently (since 1998) this category has shown more growth again. Furthermore, price effects can have caused that expenditures on foodstuffs have barely increased, while physical volumes may still have risen. Similarly, the composition within categories has often changed (see later in this chapter). The lions share of the growth in expenditures after 1950 has gone to transport (compare figure 3.3 regarding car ownership), housing (larger homes, more expensive per m 2, and diseconomies of scale due to smaller households), and other ( remainder ), which includes leisure, personal care and health. Figure 4.1 Total consumption per capita in 1995 prices (FIM) by main expenditure category food tobacco clothing housing beverages transport interior remainder Source: Laurila (1985) ( ), NA ( ), Statistics Finland; VATT Next to a simple per capita basis there exist also weighed schemes correcting for the fact that children (under a certain age) require less of some consumption items than adults do. In fact also household size is relevant due to scale economies (e.g. with respect to housing and heating). We apply here straightforward per capita comparisons (unless indicated otherwise), not the least because these weighing schemes are culture sensitive and consequently not easy to apply in a long term context.
25 17 As regards nitrogen and phosphorus flows the consumption of foodstuffs and energy is by far the most significant. For both categories it is however important to account for the context in which the consumption is realised and the consequent effects of complementary expenditures. For example, tourist or more generally leisure activities do often involve a significant amount of motorised mobility (car, motorhome, airplane). Similarly, an increased use of prepared meals and deepfreeze products implies an increased demand for refrigeration and hence for more electricity. With respect to the consumption functions to be estimated in the next phase, the complementarity issue is dealt with in terms of separability of expenditure items (e.g. Deaton and Muellbauer, 1983). As regards foodstuffs the expenditures per capita on cereals vary somewhat over the century. For meat a steady growth went on well into the seventies, since than it stabilises. On the other hand expenditures on vegetables start to become more substantial only after 1960 and this growth is still going on. A similar notion applies to fruit. In both cases it should be stressed that these products have a much higher import share than meat and cereals. The membership of the EU has enhanced this development in the nineties. Figure 4.2 Developments of expenditures per capita (in 1995 prices, FIM) for main foodstuff categories (technical trend breach ) other food sugar&sweets fruit fat & oils dairy vegetables Source: Laurila (1985) ( ); NA ( ), Statistics Finland); VATT fish meat cereals Expenditures in restaurants, which have been growing significantly since the seventies, cannot be taken into account by means of allocation to types of food. Therefore, apart from differences between volume and value developments, one should be very careful in interpreting the changing patterns in the sense of food habits due to the missing