Inevitable disaster or calculable risk?
Explorative research on the impact of future demographic development and the market for nontradable goods and services in the Netherlands
Master’s Thesis Economics August 2009 Maikel Volkerink, MSc. Student id: 0130850 Supervisors: prof. dr. C. van Ewijk
ACKNOWLEDGEMENTS
I want to thank Rob Euwals, Thijs Knaap and Leon Bettondorf for their intense support and useful comments. In particular, the private classes ‘Macroeconomics 101’ by Leon have been very much appreciated. Further thanks go out to many other people at the Netherlands Bureau for Economic Policy Analysis (CPB), like Martin Mellens, Adri Ouden, Harry ter Rele and Janneke Rijn, who always found time to be of assistance. My final thanks go out to my supervisor Casper van Ewijk, for time, comments and bulk of ideas. The views, as wells as the flaws, are those of the author and not necessarily those of the Netherlands Bureau for Economic Policy Analysis (CPB).
Index
Introduction 1
1. Unprecedented changes. Exploring and identifying future demographic and related
economic trends 5
1.1.1 Population size projections 6
1.1.2 Ageing of the population 9
1.2 Restriction at the supply side; trends in the labour market 10
1.2.1 Labour force projections 10
1.2.2 Ageing and the labour force 13
1.2.3 International comparison 16
1.3 Demographic change and future demand; trends in economic indicators 18
1.3.1 Consumption Patterns 19
1.3.2. Trade Balance, Current Account and Foreign Assets 23
2. Demand effects; future demographic changes and the relative price of nontradables 27
2.1. The model of Obstfeld and Rogoff 29
2.1.1 Defining consumption and prices 29
2.1.2 Market clearing conditions 31
2.1.3 Solving for relative prices 31
2.2 Choosing parameter values 33
2.3 The impact of demographic changes n the relative prices of nontradables 36
2.3.1 Scenarios: determining the demand quantitatively 37
2.3.2 Results: changes in (relative) prices 38
2.3.3 Preliminary conclusions 40
3. The supply side; the reallocation of labour and relative prices 42
3.1 Stylized facts about the Dutch labour market in a nontradable-tradable 42 framework
3.2 Modelling the supply side 46
3.2.1 Labour supply 46
3.2.2 The production function 48
3.2.3 Equilibrium 49
3.3 Application of the model to the case of the Netherlands 51
3.3.1 Choosing parameter values 51
3.3.2 Results 52
4. Conclusion and discussion: lessons learned 56
Literature 60
Appendixes
Appendix 1. The Dutch Household Survey and the share of (non)tradables in 63 consumption
Appendix 2. Mathematical notes to the model of Obstfeld and Rogoff 68 Appendix 3. An alternative calculations of shares of tradables in consumption (γ) 70
and production
Appendix 4. Calculations regarding the relative endowments 72
Appendix 5. The determination of the change in the price index of nontradables 73 (PN), tradables (PT) and the general price level (P)
Appendix 6. International comparison of hours worked per person per week and 76 employment rates
1
Introduction
As all European countries, the Netherlands is currently at the verge of a major transition with respect to its population size and structure. First, in contrast to developments in the previous century, the size of the Dutch population will gradually reach its upper boundary. Continuous low fertility rates (less than the replacement rate) restrict further growth of the population. In the long run, negative population growth rates are forecasted. Second, not only the size, but also the demographic make-up will alter significantly. While the share of young people falls as a result of low fertility rates, the share of elderly rises due to further increases in life expectancies. Hence, population ageing is inevitable. Whereas both trends are already in effect, they will accelerate in the coming decades and pose major challenges for the Dutch society and its economy in the (near) future.
Current economic research already exposed some of the challenges inflicted by future demographic trends. In particular, there is a vast amount of literature on the subject of ageing, both national and international.1 Mostly these studies explore the sustainability of pension schemes and
public finances in general. The main argument in this field is that current policy arrangements and institutions are inadequate in the face of future demographic developments. As the number of retirees and other economically inactive people increase relative to the amount of working people, pension benefits and other public expenditures will skyrocket to untenable levels. Economic research on population decline is scarce and fresh in comparison to the work on ageing.2 Yet, gradually a
debate is emerging and more and more economic articles on the matter of population decline are published.3
In this thesis, I want to approach the impact of future demographic trends in relation to the economy from different perspective. In my opinion, in general, economic literature neglects the structural macroeconomic changes future demographic trends bring about. The matter of ageing is neither exclusively a budgetary matter, nor will population decline solely (or foremost) affect local or regional economic dynamics. By taking the matter of demographic change out of the realm of public finance and extending the scope to a macroeconomic level, I will be able to model the impact of ageing and population decline on demand and supply. In particular, using a framework by Obstfeld
The tone of the debate, though, is rather pessimistic; focussing too much on pains and rigidities in the short- to medium run. For instance, Derks et al. (2006) anticipate structural labour shortages and increasing socioeconomic and regional economic inequalities.
and Rogoff (2004) that differentiates between tradable and nontradable goods and services4
1 Examples of research on ageing are Van Ewijk et al. (2006) and Knaap et al. (2003) for the Netherlands.
Internationally, the report by the Ageing Working Group of the European Commission (EPC, 2006), is an example of an elaborate study for the EU.
, supply
2 Likely, this can be explained by the unconventional nature of population decline. Worldwide populations have grown excessively in size during the past century. In example, between 1900 and 2000 the Dutch population size has tripled. For the majority of scholars and policy makers population decline is simply too exotic to come to mind (NIDI, 2009). 3 In the case of the Netherlands this recent attention can be explained by the fact that on sub-national level, certain regions are already experiencing negative growth rates and a declining population.
2 and demand dynamics can be analysed at a disaggregated level. This disaggregated approach is novel.5
Thinking about ageing and population decline in relation to the economy brings obvious questions to mind. How will the Dutch labour force develop? What happens with aggregated savings? Will demand for goods and services alter given an ageing population? And, if demand changes, either in composition or size, or both, are domestic production and the labour market able to adjust to the changes in aggregated demand? Börsch-Supan (2001) points to a number of mechanisms that are involved in answering these questions. Demographic change will have a major impact in all main spheres of the economy:
Typically, long-term analysis on ageing and population decline presumes that in the end all demand can be met. And if domestic supply falls short, as in the case of the Netherlands, additional goods and services are available abroad and can be paid via past current account surpluses. The novelty of my approach, in this thesis, is that by dropping the assumptions of homogeneous goods and services and perfect substitution, matters become more complicated. In particular, given those future demographic trends, substantial relative price effects may arise that cannot be neglected.
• First, the market for goods and services. With a rapidly increasing share of elderly people, demand will shift towards more goods and services specific to the elderly, like old-age and medical care. This type of shift in demand implies a structural shift of production.
• Second, the capital market. Traditionally the Netherlands is a net-exporter. As a result large stocks of capital have been accumulated abroad. As the number of retirees increases relative to working contributors, these savings will be partially repatriated to cover pension benefit expenditures. Savings are thus converted in consumption. As a result the current account and trade balance will deteriorate and run a deficit. Hence, this reversal of flows of capital also affects the market for goods and services, as demand increases at least relative to supply. • Third, the labour market. A negative growth rate of the population affects the growth rate of
the labour force. Unless labour productivity growth and participation rates skyrocket, the productive capacity of the Dutch economy becomes seriously restrained. Thus, while the amount of people of working-age falls relative to the number of retirees, is this relatively smaller workforce able to provide a sufficient level of production? And in addition, is the Dutch labour market flexible enough to match the changes in demand implied by the demography-induced changes in consumption patterns?
5 Examples of existing research on demographic change and macroeconomic issues are scare. Börsch-Supan (2001) investigates the effect of ageing and future population growth on the German labour market, though this resembles my approach only labour market consideration are considered. Another approach is offered by Lührmann (2005 and 2008), she explores the relation between ageing and consumer demand for respectively Germany and the United Kingdom. Indeed, she addresses a legitimate topic, only demand side dynamics are looked at. Knaap (2003) presents a study on the effect of ageing on wages and prices. While useful and raising important conclusions, the setup is still rather distant to reality. Finally, Atoh (2000) offers a case study of Japan (one of the first countries experiencing negative population growth rates), unable to come general conclusions.
3 Analysis of these interlinked mechanisms can be quite complex. Rather, in this thesis I will make use of a comparative static model based on a framework by Obstfeld and Rogoff (2004). Obstfeld and Rogoff are interested in the impact of an unwinding current account on the real exchange rate of the dollar. They argue that the magnitude of the adjustment of the exchange rate to the current account rebalancing is mostly determined by the level of goods market integration, as an unwinding current account implies a negative demand shock to the US economy. This is demonstrated by a general equilibrium model (of which only the demand-side is modelled) that differentiates between tradable and nontradable goods and services. The more heterogeneous good-markets are, and the more imperfect substitution is among goods and services (including international substitution), the larger changes relative prices are and in the end the appreciation of the real exchange rate of the dollar is. This two-country-two-sector approach solving for changes in relative prices will proof to be a very useful tool in analysing the impact of demographic change on the Dutch economy, as ageing and population decline can be treated into terms of demand and supply shocks as well.
My analyses are very worthwhile in relation to existing literature on future demographic change. A number of points have already been raised. First of all, the method in this paper offers a framework that discriminates between certain types of goods and services (nontradables versus tradables). Often in long term analyses, like the studies on the ageing and the sustainability of social policy (Van Ewijk et al. 2006, but many others as well), homogeneous goods and services and perfect substitution among those is assumed. Obviously, such an assumption is unrealistic. But, more
important if the relative price effects are considerable, future rates of inflation are underestimated and real rates of return overestimated. This can have serious implications for the results of these studies.
Second, and closely related, with heterogeneity among goods and services and imperfect substitution supply and demand analysis is put to another level. Given future demographic trends, there might be excess demand for certain goods and services and the Dutch trade balance may run a deficit. Optimists argue that this is no issue; the imports can be paid for by past accumulated savings. Yet, by dropping the assumption of homogenous goods and services complicates this “strategy of the current account” for the simple reason that not all goods and services can be supplied in unlimited amounts if consumers are biased towards nontradables and domestically produced tradables.
Third, my approach takes economic research on future demographics out of the realm of public finances and social policy. Ageing is not only a matter of making (future) ends meet, but also has major and enduring implications for structure of the (Dutch) economy.
Fourth, in reaction to the literature on population decline, my thesis will also reinstate the market mechanism. Structural (regional) labour shortages, as expected by Derks et al. (2006), are impossible as the market mechanism coordinates flows of labour to those places were labour is short in supply and wages are rising relative to other regions. Situations of short and excess supply are untenable in the long run. The crucial matter is whether in these new equilibriums wages and thus prices are not too high from a social or political point of view.
4 My research question is twofold:
1. Which developments and mechanisms can be identified that relate demographic change to the economy?
2. What is the impact of long-run demographic trends in the Netherlands, via these mechanisms, on the change of the (relative) price and provision of nontradables?
The structure of the thesis is as follows. Chapter 1 provides an overview of (stylized) facts and projections regarding demographic trends and related (macro)economic indicators . In discussing, papers and empirical data, main developments and mechanisms are identified to answer research question one. In chapter 2, the first part of the quantitative analyses will be presented. A comparative static model by Obstfeld and Rogoff (2004) will be introduced to estimate the magnitude of the change of the relative price of nontradables along several parametric scenarios. Chapter 2 exclusively focuses on the demand side of the economy, the structure and size of production are assumed not to alter. This latter assumption is abandoned in chapter 3. In this chapter, factors of production are allowed to change across sectors. The initial price developments of chapter 2 will be mitigated to some extend by the reallocation of labour. Adding these dynamics for the supply side to the model of O&R in chapter 2 completes the general equilibrium. Finally, chapter 4 concludes and reflects on the analyses
5
1. Unprecedented changes. Exploring and identifying future
demographic and related economic trends
Population decline and ageing will be among the main drivers of future economic development. Demographic forces have consequences for both the demand side as well as the supply side of the economy. In this chapter possible mechanisms and future trends will be explored.
With respect to demand, one path along which demographic change materialises is that of the current account. As the share of retirees increases (and accelerates), past accumulated savings (foreign assets) will be repatriated to cover pension benefits expenditures. This reversal in the flow of savings implies that the current account balance gradually deteriorates. The deteriorated current account balance, will be accompanied by a worsening trade balance. As pension benefits are used in consumption, savings are converted and the economy is faces a period of excess demand. Another mechanism that of changes consumption patterns. Elderly people are more reliant on services (like health and personal care) which possibly invokes a change in the composition of aggregated demand. On the supply side of the economy, the implications for the labour market are most profound. The labour market will be confronted with less new entrants and relatively more elderly workers. Also, to match future changes in demand for goods and services, part of the workforce has to be reallocated to other sectors of the economy. Yet, the required flexibility for such a reallocation will be restricted as a result of ageing and a shrinking workforce.
To determine the trends and discuss these mechanisms, this chapter provides an overview of academic literature and empirical research on ageing and population growth and relevant economic indicators. By discussing academic literature, anticipated trends, current figures and prognoses of demographic and economic indicators, this chapter aims to get a grip on what is ahead. In addition, these empirics and insights will be used in determining the values of the parameters in the models of chapters 2 and 3. In particular, there will be a focus on developments in terms of nontradables versus tradables as this is a central aspect of the quantitative part of this thesis. Note that projections are surrounded by large uncertainties. The outcome of forecasts is determined by the methodology of the underlying studies. For this reason, if possible and relevant, multiple projections by various scholars and various institutions are presented for each indicator. In a number of cases underlying
assumptions are even explicitly discussed.
The remainder of this chapter is divided in three parts, split in several sections. Part one covers demographic indicators on population size and ageing. Part two deals with developments related to the supply side of the economy, the labour force, exploring future participation rates and projections on the labour force size. The third part addresses demand and focuses on the potential impact of demographic forces on patterns of consumer spending and flows of capital. In the end, by this approach all relevant developments in the labour market, capital market and the market for goods and services are explored.
6
1.1.1 Population size projections
Figure 1.1 depicts historical values and prognoses for the Dutch population. As is evident from the chart, future demographic developments will be very different from those in the past. Between the year 1900 and 2000, Dutch population growth was substantial, tripling from 5,1 to 15,8 million people. Currently, the size of the Dutch population is reported to be around 16,5 million (CBS, 2009). In their projections, both Statistics Netherlands and the United Nations Population Division expect the Dutch population to reach its ceiling in the coming decades, peaking in 2037 or 2038 at an estimated size between 17,0 million (UN Pop, 2006) and 17,5 million (CBS, 2008a). During the 2040s a slight decline is expected to set in.
Figure 1.1 Dutch population. Total and by age-group. Historical and projections
Source: CBS (2009) and UN Pop (2006)
In addition to total population dynamics, figure 1.1 also includes values for subgroups. It shows that the decline is most apparent (and already started) among the young (aged 0 to 19 years) and working-age (20 to 64 years) groups. The number of elderly (aged 65 years and older) continues to grow in the next 20 to 30 years and is expected to double in size from 2,3 million in 2007 to 4,3 million at its peak in the late 2030s.
0 2 4 6 8 10 12 14 16 18 1900 1920 1940 1960 1980 2000 2020 2040 Pop (mln) Year
7 Table 1.1 Historical and projected figures on key population indicators
Population size (x1.000) 1900 1950 1980 2000 2020 2030 2040 2050
CBS, base projection 5.104 10.027 14.091 15.864 17.014 17.380 17.474 17.343
Scenario Global Economy* 17.884 18.890 19.677 20.335
Scenario Strong Europe* 17.574 18.325 18.852 19.234
Scenario Transatlantic Market* 16.961 17.165 17.104 16.836 Scenario Regional Communities* 16.500 16.334 15.846 15.107 Fertility rate 4,45 3,10 1,60 1,72 1,75 1,75 1,75 1,75
Net migration (x1.000), CBS 20 53 54 14 14 15 15
Net migration (x1.000), Carone 33 32 31 31
Life expectancy at birth 71,6 75,9 78,0 81,0 81,7 82,3 82,8 Source: All CBS (2009), except additional population scenarios by De Jong and Hilderink (2004), marked "*" and migration figures by Carone (2005)
Developments in population growth are the result of three factors: birth (total fertility rate), death (life expectancy) and net migration. Table 1.1 presents figures about these variables for a selected number of years. A continuous low fertility rate seems an obvious determinant for the decline in population growth. In 1900, on average, women gave birth to more than four children (TFR = 4,45). Currently the rate is about 1,72, well below the replacement rate – the rate to sustain a population, of 2,07 (Beets, 2008). This drop in the fertility rate might be offset by increasing life expectancies and positive net migration. Indeed, life expectancies at birth are expected to increase further in the future, yet this will not yield additional Dutch citizens. It will only raise the share of elderly people in the population – ageing. Second, migration. If enough people move to the Netherlands, the population increases in size. Again, as follows from table 1.1 positive net migration is prognosed. Nevertheless, the amount of migrants will be too low and only mitigate the effect of low fertility rates to a limited extend.
All in all, Statistics Netherlands (CBS, 2009), as well as international studies, anticipate that population growth in the Netherlands will come to a halt in the coming decades. However, despite the consensus among scholars, the prognoses are far from rock-solid. First, a simple glance at historical figures on migration to the Netherlands reveals that past flows are very volatile (Figure 1.2). This makes any projection on the number of migrants very uncertain. De Jong and Hilderdink (2004) make the uncertainty involved in forecasting even more explicit. Instead of taking key determinants of population growth (fertility rate, life expectancy and net migration) as exogenous, they assess how these determinants are affected by future socioeconomic and socio-cultural
developments. For instance, on the individual level the fertility rate is the outcome of microeconomic decision-making based on income, level of education and availability of day-care facilities. Drivers of net-migration are, for instance, the rate of economic growth and the degree of openness of national borders and labour market. By outlining four very different scenarios on the drivers of population
8 Figure 1.2 Historical path of net migration
Source: CBS (2009)
Figure 1.3 Dutch population, multiple projections
Note: GE, SE, RC and TM all refer to the scenarios applied by De Jong and Hilderdink (2004), see footnote 6 Source: CBS (2009), UN Pop (2006) and De Jong and Hilderdink (2004)
-60 -40 -20 0 20 40 60 80 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Year Migrants (x1.000) 14 15 16 17 18 19 20 21 2010 2015 2020 2025 2030 2035 2040 2045 2050 Pop (mln) Year GE SE TM RC CBS UNPOP
9 growth for Europe, De Jong and Hilderdink map four distinct trajectories for the Dutch population.6
In sum, given the current low fertility levels and the unlikeliness of a substantial rise of this rate in the future together with a limited inflow of migrants, Dutch population growth will come to a halt in the coming decades. Though the exact path and level is debatable and dependent on a broader array of socioeconomic and socio-cultural developments as well, as De Jong and Hilderdink (2004) have shown, in this thesis the prognoses of Statistics Netherlands is followed. This is also pragmatic; many other projections, to be discussed, incorporate these forecasts as well.
These distinct trajectories are included in table 1.1 and charted in figure 1.3. In the most favourable case (Global Economy, GE), population growth persists and the Dutch population increases to 21 million by 2050 and continues to grow thereafter. An opposite scenario (Regional Communities, RC), however, results in an early and persistent negative growth. These differences in the development of population growth are mainly caused by changes in net migration. The GE scenario presupposes increased openness of the Dutch economy and labour market plus high potential GDP growth. Together these conditions invoke a large flow of migrants. In addition, these conditions foster a high fertility rate. Regional and economic outlooks are more restricted in the RC scenario.
1.1.2 Ageing of the population
Declining population growth rates pose not the only demographic challenge. The demographic make-up will alter as well. As trends in fertility rates (low) and life expectancy (increasing) are persistent in the long run, the share of elderly in the total population will continue to increase. Figure 1.4 gives an impression of the trend in ageing in terms of the old-age dependency ratio (OAD-ratio). This ratio represents the number of elderly (defined as aged 65 and up) relative to the number of working age people (defined either as aged 15 or 20 to 64).
Currently the ratio is in the range 20-25% (depending on the definition of the working age population). As is apparent from figure 1.4, in the coming decades this ratio rapidly increases and eventually doubles. At its peak around 2040, the OAD-ratio is estimated at about 45%. During the 2040s a gradual decline sets in. As the OAD-ratio increases there will be a larger number of elderly relative to the number of people working age. In effect, this implies that the number of people active in the labour force will have to support more elderly (and other economically inactive citizens). This burden has many implications, not only in terms of public finances, but also in real economic terms. Can this relatively smaller amount of people produce sufficient goods to match the demands of the whole population?
6 De Jong and Hilderdink base their calculations on four trajectories for Europe from the CPB study by De Mooij and Tang (2003). De Mooij and Tang create these scenario’s along two dimensions: (i) private versus public
responsibilities and (ii) international cooperation vs. national sovereignty. This results in the following: Global Economy (GE), Strong Europe (SE), Transatlantic Market (TM) and Regional Communities (RC). Further elaboration on these scenarios is beyond the scope of this thesis.
10 Figure 1.4 Old-age dependency ratios, multiple projections
Source: Carone (2005); CBS (2009); Van Ewijk et al. (2006); Knaap & Bettendorf (2003); UN Pop (2006)
1.2 Restrictions at the supply side; trends in the labour market
As said in the introduction, the demographic changes ahead will confront the labour market with serious challenges. Reduced population growth and a declining number of people of working age (aged 20 to 64) will seriously limit the amount of labour supplied and thereby restraining productive capacity. Section 1.2.1 presents projections on the size of the labour force. In addition, population ageing also implies an ageing working force. Ageing will have an impact on the flexibility of the labour market. Though older people are more experienced, they are also less mobile. With an ageing labour force future sectoral adjustments thus may be more difficult (section 1.2.2).
1.2.1 Labour force projections
Figure 1.1 showed that the size of the potential labour force (those aged 20-64 years) is already declining. Though developments of the potential labour force are essential, it is important to note that the trajectory of the labour force does not need to follow demographic trends one-to-one. The path of the labour force is also determined by developments in the labour force participation rate.7
7 The labour force participation rate is defined as the percentage of the potential labour force (in most studies that is all people in aged 15 or 20 to 65) that either has a job or is actively looking for a job. Statistics Netherlands defines the potential labour force as all people aged 20 to 65 years. The labour force participation rate is defined as the share of this potential labour force that works, or is willing to work, for at least 12 hours a week. Internationally, for example
0 5 10 15 20 25 30 35 40 45 50 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Ratio (%)
11 Theoretically it is possible to maintain a constant size of the labour force with a declining population as long as this is compensated by an increasing participation rate. It is thus important to know how the participation rate develops in the future.
In projecting the participation rate Euwals and Van Vuuren (2005) consider three factors. First, a demographic composition effect. Observations confirm that for different age categories different rates of participation exist. Generally, elderly workers (aged 55 and above) have low levels of participation. Young people (aged minus 25) participate even less. Depending on the relative sizes and growth rates of these age categories, overall participation increases or declines. Second, women participate more and more. One source is a cohort effect. Old, less participating, cohorts are replaced by younger, more participating, cohorts. In addition, there are socio-cultural developments that foster persistent growth of female participation. Finally, participation rates are also influenced by previously implemented (social security) policies, like changes in (early-) retirement policies and disability arrangements.
Euwals and Van Vuuren find that participation will increase in the future, but large increases as in the past seem unlikely. This development is seconded in a comparable study by Carone (2005). The figures are presented in table 1.2.8 Indeed during the 1990s a substantial gain in labour force
participation was realised, however for the coming decades increases in participation are minimal. Looking at the developments by gender, figures for male and female participation reveal that future increases in participation are on account of females. Male participation rates are even expected to decline slightly. Carone offers another dimension by including participation rates for age-groups. His figures show that next to increased female-participation, future percentage-points increases are also partially the result from to a catch-up in participation by elderly workers (aged 50 and up).
Table 1.2 Participation rates in the Netherlands, multiple projections
Euwals and Van Vuuren (2005) 1990 1995 2000 2005 2010 2020 2030 2040 2050
Total 63,0% 67,0% 71,0% 72,0% 73,0% 76,0% 74,0% 75,0% 74,0%
Male n/a n/a 84,0% n/a 81,0% 82,0% n/a 82,0% n/a
Female n/a n/a 58,0% n/a 65,0% 69,0% n/a 68,0% n/a
Carone (2005)
Total 77,1% 77,8% 79,1% 79,3% 80,8% 80,5%
Male 84,0% 83,1% 82,8% 82,3% 83,5% 83,2%
Female 70,1% 72,4% 75,4% 76,2% 77,9% 77,6%
Elderly workers 47,5% 49,0% 54,1% 53,8% 55,4% 56,0%
Source: Euwals and Van Vuuren (2005); Carone (2005)
in the studies by the European Commission (Carone, 2005), this threshold of 12 hours is dropped and broader share of the population (aged 15 to 65) is included.
8 In both table 1.2 and figure 1.5, estimates in participation and size of the labour force differ due to differences in definitions of the labour force. Euwals and Van Vuuren use a more restricted definition of the labour force (equal to that of Statistics Netherland, see also footnote 5). For this reason Carone reports both a larger size of the labour force and a higher rate of participation.
12 Figure 1.5 Labour force projections (including trendline)
Source: Euwals and Van Vuuren (2005); Carone (2005)
Having established the trends for labour force participation (a slight increase) and the population size (zero-growth and possible decline), the size of the Dutch labour force can be calculated as a product of these two figures. Both Euwals and Van Vuuren (2005) and Carone (2005) have made projections, which are presented in figure 1.5. All authors project the same trend in labour supply: a limited increase between 2005 and 2020 (no more than half a million), a larger decline between 2020 and 2035 (with the labour force size even falling below its current size) and finally a gradual restoration to the current level during the 2040s. Again, the difference between the projected sizes of the labour force is the result of slight differences in definitions (see also footnote 9).
In sum, the trajectory of the size of the labour force follows to some extent that of the working-age population, however growth rates of the labour force are partially mitigated by a slight increase of the labour force participation rate. Still, demographic forces are dominating. While the number of working-age people (the labour force) is already declining, the fall in the size of the labour force sets is in a little later (around 2020). The size of the labour force is smallest during the heyday of ageing in the 2030s. Note, that future trends in labour supply, that is the number of hours people are willing to work, are not considered. These are not projected. For the Netherlands it is known that the number of hours worked per week is relatively small in international comparisons (Alesina et al., 2005, see appendix 6). This is mainly due to a large share of part-time workers. Another remark, as will be discussed in chapter 3, is that the average number of hours worked differs across sectors. On average, employees in the financial and commercial services sectors work more hours a week than workers employed in the health and long-term care sectors. Hence, if labour supply (in terms of volume of hours) increases substantially, a fall in total labour supply might even be prevented. Here there is room for government policy. Yet, most likely, to increase the number of hours worked wages have to increase substantially, venting inflation.
6 6,5 7 7,5 8 8,5 9 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Labour force (mln) Year
13
1.2.2 Ageing and the labour force
As the population ages, so will the labour force. Elderly workers have both advantages (like more experience) as well as disadvantages (less flexibility). A rapidly ageing workforce can therefore have major implications for a country’s productive capacity. In addition, it can lead to profound personnel problems if a too large part of the workforce in a company or sector is relatively old and soon retires. A more detailed look at the matter of ageing in the workforce is thus very relevant for the purposes of this thesis. A high degree of ageing and/or large differences between sectors (particularly between the nontradables and tradables sector) may be a sign of low levels of flexibility at the labour market. An observation that is useful for the analyses in chapter 3. Based on data from the Dutch Labour Force Survey (EBB) several informative graphs are constructed that address the issue of ageing and the labour force.
Table 1.3 displays the average age and the generation index for various sectors for the years 2000 and 2007. Table 1.3 highlights that average age varies a lot across sectors. In 2007 the average age in the Dutch labour force was 41,4 years. Personnel in the education sector was the oldest with an average of 45 years. Staff in the hotel, restaurant and catering sector (37,2 years) and retail trade (38,7 years) are relatively young. Temporary workers are identified separately.9 Indeed, they are
very young (32,9 on average), supporting the argument that flexibility is a feature of young workers. It is also possible to classify the tradables and nontradables sector.10 Perhaps contrary to common
expectations, on average the labour force in the nontradables sector is younger than in the tradables sector. Yet, the difference is small; only one year. By including data for the year 2000 it is possible to a get glimpse of the evolution of ageing over time. During the period 2000-2007 overall the average age increased by 2,4 years. The pace of ageing is not equal across sectors as well. In the agricultural sector and the energy and water sector the average age increase was less than one year, whereas in manufacturing sector and the transportation and communication sector the increase was close to three years.11
In addition, table 1.3 also includes the socalled generation index (Bruggink, 2008). This index represents the ratio of persons of age 50 in the workforce relative those aged 30. A value below one thus implies that there are relatively more 30-years old workers, a value above unity means the opposite. An increasing ratio thus give some sense of the pace of ageing. Table 1.3 shows that in 2000 the overall index for the Dutch labour force was 0,57. In 2007 the index had already increased to 0,84. In comparison to the average age, the index gives comparable, but slightly, different insights.
9 On account of Bruggink (2008) I have excluded temporary personnel from the analysis as these are very young and disproportionally represented across the sectors. Including them would distort the structural image of the sectors. 10 The method of classification will extensively discussed in section 2.2 and appendix 3.
11 On account of the energy sector, the low increase seems to be caused by the liberalisation of this sector. In preparation of the liberalisation external advisors were temporarily hired (typically young people) and old workers were offered to leave their job before the retiring-age (Bruggink. 2008: 8). In the agricultural sector the low increase is most likely related to the already high average age.
14 Table 1.3 Average age and generation index per sector
Average age Generation index 2000 2007 Difference 2000 2007 Difference Education 43,6 45,0 1,4 1,43 1,91 0,48
Energy and water 43,4 43,8 0,4
1,22 1,36 0,13
Mining 42,5 43,5 1,0
0,43 0,94 0,51
Agriculture, forestry and fishing 42,2 43,0 0,8 0,84 0,92 0,08 Public services 41,2 42,9 1,7 0,71 1,13 0,42 Manufacturing 39,2 42,2 3,0 0,55 0,83 0,28 Health care 39,6 42,2 2,6 0,56 1,11 0,55 Transportation, communication 38,7 41,6 2,9 0,49 0,85 0,37 Other services 39,3 41,2 1,9 0,72 0,81 0,09 Construction 37,4 40,3 2,9 0,57 0,82 0,25 Financial services 38,4 40,2 1,8 0,45 0,48 0,03 Commercial services 37,7 40,1 2,4 0,40 0,50 0,10 Whole sale, retail trade and repairs 36,5 38,7 2,3
0,47 0,61 0,14 Hotel, restaurants and catering 35,8 37,2 1,4
0,46 0,59 0,13 Temporary and flex workers 32,8 32,9 0,1
0,44 0,68 0,24 Tradables sector 39,5 42,1 2,6 0,57 0,84 0,28 Nontradables sector 38,8 41,1 2,3 0,58 0,85 0,27 Overall 39,0 41,4 2,4 0,58 0,85 0,27 Source: own calculations, based on EBB (2000) and EBB (2007)
Indeed, for the education sector the degree of ageing is again highest (the index is 1,91 in 2007). And again, the hotel, restaurant and catering sector has a low index value (0,59). However, the generation index is smallest for the commercial and financial services sectors (both close to 0,50). So while the average age in these sectors is not the lowest, there are many 30’ers employed relative to the 50’ers. Second, the generation index reveals that the pace of ageing is highest in the health care sector with an increase of 0,55 between 2000 and 2007. This is in line with the high increase in average age for this sector. Finally, in terms of tradables versus nontradables the generation index displays no substantial differences. Both for 2000 and 2007 the indexes are (almost) alike and on average.
Finally, figure 1.6 might be the most informative. It shows the make-up of the labour force (total and per sector) per age-category. Again we find that the education sector may face the most serious problems in terms of ageing. Already in 2007 over half of the staff was 45 years or older and almost 20% over 55 years. This implies that within two decades more than half of the current staff retires. Given the demographic projections, it is than questionable whether this sector is able to maintain sufficient staff. Based on figure 1.6, sectors like construction, retail trade and commercial services seem better of. Last, the shares of the older age categories seem to be slightly bigger in the tradables sector in comparison to the sizes of the shares for nontradables sector. Yet, again the differences are rather small.
15 Figure 1.6 Dutch labour force (2007), per age group, per sector
Source: own calculations, based on EBB (2007)
All in all, table 1.3 and figure 1.6 provide some rough conclusions on current conditions and a taste of what is ahead. First of all, both figures reveal that the degree or level of ageing is varies a lot across sectors. Second, the figures did not provide a clear image of whether ageing is more severe in the tradables or nontradables sector. Though data does suggest that the tradables sector is a bit worse off, that is more aged, difference are too small to make bold statements. In fact, both the nontradables sector and the tradables sector include subsectors that score above-average in terms of ageing (like education in the nontradables sector) as well as subsectors that rank below average (like financial services). Third and final, though an international benchmark is lacking it seems that the Dutch labour force is ageing rapidly. In particular the education sector, and to some extent the health care sector and public services, are highly aged. Detailed analyses of development of the labour force (like study choice and sizes) are welcome to assess future bottle-necks.12 Moreover, an aged labour force
may be a problem as it is observed that job-to-job and intersectoral mobility is smaller among older workers in comparison to the young workers.13
12 More elaborate studies on sectoral developments in the future are CWI (2007) for the short-run and Huizinga and Smid (2004) for long-run scenarios. A study by the ROA (2007) is an excellent example of the matching of graduates and vacancies in the medium-run.
Future demographic developments may thus
13 Regarding the issue of labour mobility in relation to age see for instance: Schumacher (2007), Light (2005) and Muhleisen and Zimmerman (1994).
0% 20% 40% 60% 80% 100%
Average Tradables sector Nontradables sector Education Agriculture, forestry and fishing Mining Energy and water Public services Health care Manufacturing Transportation, communication Other services Construction Whole sale, retail trade and repairs Commercial services Hotel, restaurants and catering Financial services Temporary and flex workers
15 to 24 year 25 to 34 years 35 to 44 years 45 to 54 years 55 to 64 years
16 restrain flexibility in the labour market. A serious matter, especially if the same demographic forces invoke changes in the composition and size of consumption and a reallocation of labour is required. In the past, most of the gaps during these periods of reallocation were filled by the new entrants. Yet, as the Dutch the labour force is expected not to grow in the coming decades, and even falls in the medium run, future sectoral readjustment will depend foremost on the degree intersectoral labour mobility. This issue is taken up again in chapter 3.
1.2.3 International comparison
The previous sections dealt exclusively with the Netherlands. Yet, we cannot regard the Dutch case in isolation from the rest of the world. Hence, before moving to the economic and financial indicators, this section concludes with a discussion of demographic trends in selected other countries and regions.
Figure 1.7 presents projections on average population growth per 5-year period for the world, Europe, Germany and the Netherlands.14 Although Dutch growth rates are lower than world
population growth rates, it turns out that in a European context the Netherlands is performing well: population decline is limited and sets in late. Germany suffers from a far worse problem. Part of an explanation could be that the fertility rate in the Netherlands is still relatively high in comparison to other European nations (UN Pop, 2006).
Figure 1.7 Comparison of average projected population growth rates (%)
Source: UN Pop (2006)
14 In this section it is chosen to focus on Germany in particular, as it is a major trading partner of the Netherlands. However, in a European context, demographic developments are not most severe for Germany. The current situation and projections are worse for Italy. (Carone, 2005).
-0,6% -0,2% 0,2% 0,6% 1,0% 1,4% Period World Europe Germany Netherlands
17 Figure 1.8 Comparison of projected OAD-ratio’s (%) and shares population 65+ (%)
Source: OAD-ratios, Carone (2005); Share 65+, UN Pop (2006)
Figure 1.9 Comparison of average projected labour force growth rates (%)
Source: own calculations derived by recalculating projected absolute sizes of the labour force by Carone (2005) 0% 10% 20% 30% 40% 50% 60% 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Year
65+ World 65+ Europe 65+ GER 65+ NED
OAD World OAD EU-15 OAD GER OAD NED
-5% -3% -1% 1% 3% 5% Period EU25 EU15 Germany Netherlands
18 Second, in terms of ageing, the Netherlands performs close to the European average as is illustrated by figure 1.8. The United Nations Population Division (UN Pop, 2006) has prognoses on the share of the population of age 65 and up. In Germany the share of elderly is highest during the whole period. The Dutch share deviates little from the European average. Comparison of the OAD-ratios by Carone (2005) provides a similar image. Carone has figures for EU-members and EU-totals, of which figures for the Netherlands, Germany, EU-25 and the world are included in figure 1.8 too. Finally, note that after the peak around 2040, during the remainder of the 2040s, both the OAD-ratio and the share of population aged 65+ falls in the Netherlands, whereas for the other selected countries and regions both indicators continue to increase.
Finally, what are the international projections for labour force sizes? Carone (2005) not only projects labour supply for the Netherlands, but for all EU-members and EU-totals. Figure 1.9 shows 5-year period average growth rates. Developments of the labour force mirror those of the population. In Germany the decline sets in earlier, is largest and most persistent. The period of negative labour supply growth is shorter and less severe for the Netherlands, even in comparison to EU-averages.
This brief scan of international demographic developments support the claim that the demographic developments in the Netherlands are not unique. In the rest of the world there are comparable trends. Yet, on all indicators it seems as if other countries or regions (like Germany or the EU-25/Europe) are facing bigger problems: population decline is more severe and the degree of ageing is higher. For the Netherlands this observation makes future strategies of importing all excess demand or hiring additional labour on the international labour market naive and not viable. Simply, there are many other countries experiencing bigger demographic pressures which are competing for the same scarce resources.
1.3 Demographic change and future demand; trends in economic indicators
Restricted growth and ageing not only affect the supply side of the economy by limiting the potential of the labour force. Demographic change has an impact on the demand for goods and services as well. First, demographic trends may alter the composition of aggregated demand. Consumer preferences are likely to be different for people of different ages. So, if preferences for goods and services differ a lot with age, the make-up of aggregated demand may alter substantially as the share of retirees in the population increases. This induces relative price effects. Second, in terms of flows of capital, soon a considerable sum of national savings will be repatriated as pension benefits need to be paid. Hence, the Dutch foreign asset position deteriorates and the current account balance shifts from a surplus to a deficit. At the same time this money will be used in consumption, causing domestic consumption to increase relative to domestic production. The deficit of the current account will thus be accompanied by a deficit of the trade balance. This period of excess demand is another channel at the demand side of the economy along which relative prices may be affected.19 Together, these two mechanisms (and related indicators) in the market for goods and services and the capital market embody what can be regarded as the demography-induced “demand shock” in the forthcoming analyses of chapter 2. In this final section both mechanisms will be explored. In
particular, these channels will be related to the nontradables-tradables framework.
Recall, nontradables should be thought of as goods and services that can only be produced domestically and supplied to the home market. Hence, nontradables cannot cross borders, tradables can. This restriction either follows from characteristics of the good itself, like a haircut or nursing, or because of high transport costs or regulation, like tariffs. Contrary to other long-run scenario studies (like Van Ewijk et al., 2006) in this thesis homogeneous goods and services and perfect substitution are not assumed. Presuming heterogeneous goods and services complicates equilibrium analysis and prognoses. For instance, if it turns out that demand for health care (a nontradable) increases, future excess demand cannot simply be met by additional imports. Instead, either the supply of health care should be rationed, or the price of such services should rise, or productive capacity in the health care should increase at the cost of production in other sectors (as growth in labour supply is limited). Irrespective of the outcome these dynamics cannot be neglected. Moreover, by assuming imperfect substitution between tradables, strategies of specialising in nontradables production and importing the rest are not optimal; domestic consumers may prefer domestically produced tradables over those produced abroad. These issues will be extensively dealt with in chapters 2 and 3. At this point, these issues justify the relevance of exploring present and future consumption patterns and expected trends in flows of capital.
1.3.1 Consumption patterns
Population ageing does not leave consumption unaffected. Common knowledge suggests that elderly spend a relatively larger sum of their income on personal and medical care and less on education or transportation. If this is true and changes in those shares are considerable, ageing will cause a shift in aggregated demand towards goods and services that elderly most prefer. Changes in the composition of consumer spending is thus one channel along which the effect of ageing on demand materialises.
The relation between age and consumption patterns is still indefinite, mostly since empirical research on the expenditure patterns of age-groups is still scarce. Börsch-Supan (2001, in a study on ageing and the labour market) takes a brief look at the relation between age and consumer spending. By comparing budget shares for different consumption categories per age-group, he establishes that the budget shares are different across age. However, Börsch-Supan remarks that as he uses cross-sectional data, he is unable to disentangle age, time and cohort effects.15
15 The age-effect is the relation between a certain behaviour and age; this is assumed to be fixed over time. The cohort-effect implies that differences in behaviour (like consumer spending) are specific to the cohort. In contrast to the age-effect this effect is not constant over time: a 60-year old in 2007 acts differently than a 60-year old in 2027. Finally the time-effect refers to behaviour that is specific to a point in time: an example is the effect of the “flower power era” on preferences of all people during that period.
20 Table 1.4 Relative importance weights of major expenditure categories
Category CPI-W CPI-E
Food 17,91 14,32 Housing 37,64 45,94 Apparel 3,97 2,77 Transportation 18,89 13,81 Medical care 5,06 10,24 Recreation 5,34 4,36 Education 5,09 2,98 Other 6,09 5,59
Source: Hobijn and Lagokos (2003:2)
that age is the main determinant of the changes in spending behaviour. Repeated cross-sectional analysis showed that the age-consumption profiles did not change much across age-groups during the 1980s and 1990s, hence the influence of cohort and time effects are dominated by the age-effect.
Hobijn and Lagakos (2003) come to comparable conclusions, though their point of departure is different. Hobijn and Lagakos criticise the consumer price index (CPI) that is used to index retirees’ benefits. The CPI used, captures price changes in the average set of goods and services purchased by urban workers. Retirees, they argue, allocate their purchases differently. Hence, the CPI (based on behaviour of urban workers) is incorrect to index pension benefits. Their results are presented in table 1.4. Clearly, the weight attached to medical care by the elderly is twice as large as that of the urban workers. Purchases on transportation fall by a third. Hobijn and Lagakos conclude that given these differences in consumption patterns, inflation is higher for retirees than for urban workers. Hence, the purchase power of the elderly is structurally overestimated.
The previous two studies give empirical support to the claim that consumer spending is different for people of different age. Yet, for projections it is important whether these differences are due to age, cohort or time effects or a combination. Though panel data is not available, Lührmann (2005 and 2008) has looked more carefully at this issue. She concludes that most of the difference is caused by the age-effect; cohort- and time-effect contribute little. Current budget shares are a good proxy for future budget shares and can safely be used in projections.16
Figure 1.10 gives patterns of consumer spending per age group for the Netherlands.
17
relatively more expensive (see for instance Börsch-Supan, 1992). Second, the chart also echoes the Based on Dutch household data it gives the budget shares for seven main categories of goods and services. The age of the household is linked to the age of main earner. While the categories are a bit rough some trends are clear. Moreover, the trends portrayed in this figure are comparable to the findings of Börsch-Supan (2001). First of all, for the elderly the consumption share on housing is considerable. Often it is suggested that is due to a drop in income during retirement, while the costs of housing remain the same as elderly generally do not move to smaller cheaper apartments. Housing becomes
16 Which is exactly what is done in the upcoming analyses for figures 1.10 to 1.12. Current budget are kept constant in the future without correcting for cohort- or time-effects.
21 Figure 1.10 Budget shares private consumer spending per age group
Source: own calculation based on CBS Budget Onderzoek (2000)
Figure 1.11 (Projected) Budget shares private consumer spending, 7 categories
Source: own calculation based on CBS Budget Onderzoek (2000) and CBS (2009). See appendix 1.
18% 18% 18% 20% 19% 18% 17% 22% 22% 22% 21% 26% 28% 33% 8% 8% 8% 7% 8% 8% 7% 5% 6% 6% 6% 7% 7% 8% 15% 12% 13% 15% 13% 12% 10% 16% 14% 14% 14% 11% 13% 9% 17% 20% 19% 17% 17% 15% 16% 0% 20% 40% 60% 80% 100% 20-24 25-34 35-44 45-54 55-64 65-74 75+ Age group Other goods
Transportation and communication Education, entertainment and leisure Health and personal hygeine Fuel, gas and electricity Housing
Food, drinks, tabacco
19% 18% 18% 18% 18% 22% 29% 24% 24% 25% 8% 8% 8% 8% 8% 6% 7% 6% 6% 6% 13% 11% 13% 13% 13% 14% 12% 13% 13% 13% 19% 15% 18% 18% 17% 0% 20% 40% 60% 80% 100%
Working Age Retiree Population Average (2000) Population Average (2020) Population Average (2040) Other goods
Transportation and communication Education, entertainment and leisure Health and personal hygeine Fuel, gas and electricity Housing
22 Figure 1.12 (Projected) Budget shares total consumption, tradables versus nontradables
Source: own calculation based on CBS Budget Onderzoek (2000), Van Ewijk et al. (2006) and CBS (2009). See appendix 1.
intuition that private consumer spending on health and related goods and services increase with age. In contrary, consumer spending on transport and communication, as well as foods and drinks, falls with age. Together these trends and an anticipated rapidly ageing population, justify the expectation that shares of aggregated demand of housing, health and personal hygiene should increase in the coming decades.
Based on the findings in figure 1.10 (and extended analyses and calculations in appendix 1) figures 1.11 and 1.12 present estimations on the composition of consumption in the future. Keeping the current budget shares of the two age groups (working age and retirees, in blue) constant over the future, it is now possible via current and projected dependency ratios (by Statistics Netherlands) to determine ‘national’ or ‘population average’ budget shares for 2000 and in the future (the grey bars).18
Figure 1.12 translates the analyses into the nontradables-tradables framework. In addition, public expenditures are included (this was impossible at the level of disaggregation in figures 1.10 and 1.11). The blue bars show, that the share of nontradables consumption is considerably larger for retirees (61,2%) in comparison to those people of working age (45,4%), Still, while keeping these shares constant in the future and projecting the overall composition of consumption, again the effect of demographic change is limited (the grey bars): the share of nontradables only increases a few
Figure 1.11 shows future developments for private consumer spending, solely based on the data underlying figure 1.10. A comparison of the stacked bars for the years 2000, 2020 and 2040 may be disappointing. Even though the number of elderly in the Netherlands and the dependency ratio double, changes in the overall budget shares are limited.
18 Again, see appendix 1 for details.
45,4% 61,2% 48,2% 49,4% 50,6% 54,6% 38,8% 51,8% 50,6% 49,4% 0% 20% 40% 60% 80% 100%
Working Age Retiree Population Average (2000) Population Average (2020) Population Average (2040) Tradables Nontradables
23 percentage points. All in all, it seems as if the impact of ageing on the composition of demand is not profound. Yet, such a conclusion may be misleading. Indeed, on the presented levels of aggregation (seven or two categories) no apparent changes in demand are shown. However, note that at these levels substitution within the categories are concealed. Take transportation and communication. When spending on transportation goods and services falls, while consumption of telecommunication goods and services increases, the net change in the share of this category in total consumption is minimal. Hence, further disaggregation of the categories possibly yields larger changes. Second, in figures 1.10 and 1.11 only private consumption is displayed, public expenditures are not included. Especially since public expenditures on health care are high, including public expenditures might already yield larger changes in the budget shares.19 An option that is not excluded by the limited
changes in figure 1.12, as the bundles nontradables and tradables also conceal changes at more disaggregated levels.
1.3.2 Trade Balance, Current Account and Foreign Assets
Demographic forces do not only affect the demand side of the economy by altering the composition of demand. As the number of retirees increases relative to the working-age population, relative demand (to supply) increases as well. The logic is quite intuitive. With an increasing number of retirees, past savings have to be addressed to cover all pension benefit expenses. Hence, the Dutch foreign assets position declines and the current account deteriorates. At the same time this reverse flow of savings will be converted into consumption by the retirees. Consumption (demand) outgrows production (supply) and additional demand will be met by increasing imports (to exports). The trade balance thus deteriorates as well. Obviously, such shifts in assets positions and changes in current account and trade balances will affect relative prices. The magnitude of these adjustments depend on the flexibility of goods-market (as stressed by Obstfeld and Rogoff, 2004). Chapters 2 and 3 deal with the magnitudes of these adjustments. This section merely provides an overview of anticipated future developments of the trade balance, current account and the NFA-position for the Netherlands.
Historically the Netherlands has been a net-exporter running a positive trade balance. In addition, though net factor income and transfers varied, these have never been large enough to offset the trade balance. As a result, the Netherlands ran sizable current account surpluses in the past; each and every year national savings accumulated. It is this foreign asset position that is sometimes considered for paying future trade deficits and smooth consumption. That is, the strategy is plausible as long as these deficits are not too large and too persistent. This brings up the question how the trade balance and current account develop and whether the Dutch foreign assets position is large enough?
19 Public expenditures can unfortunately not be included in figures 1.10 and 1.11 as data on public consumption is poorly specified. Public expenditures are included in the shares in figure 1.12. Again, see appendix 1 for a discussion.
24 Table 1.5 Projections of Trade Balance and Current Account (% GDP) and NFA-position
Bettendorf and Knaap (2003) 2006 2011 2020 2030 2040 2050 2060
Current Account 5,6% 6,2% 5,6% 4% -1,8% -1,75% 0,3%
Trade Balance 4% 3,3% 1% -2,3% -6,9% -4,8% -2%
NFA-position (% of GDP) 55% 75% 120% 145% 125% 80% 60%
Van Ewijk et al. (2006)
Current Account 8,2% 4,7% 6,2% -1,9% 0,4%
Trade Balance 9,4% 5,3% 4,5% -4,9% -2%
NFA-position (index: 2006 = 100) 100 176,1 203,1 186,3
Source: Bettendorf and Knaap (2003), E: Van Ewijk et al. (2006)
Van Ewijk et al. (2006) and Bettendorf and Knaap (2003) have modelled the future path of the trade balance and the current account, in view of an ageing population. The projections can be found in table 1.5. Both balances in both projections display a similar trend. The trade balance will be in surplus in the coming decade, runs a deficit afterwards, but after the peak around 2040 the deficit declines and is expected to move into a surplus again. The current account follows a similar path, yet the deficit will be smaller and of shorter length, as positive net factor and transfers balances offset the trade deficit. Indeed, continued consumption by retirees will cause a period of excess demand and force the Netherlands to shift from a net-exporter to a net-importer position. Yet, if the stock of past and coming current account surpluses is large enough, that is when net savings are sufficient, the Dutch population should be able to pay for this temporary period of deficits.
Figure 1.13 presents the historical development of the current account and its accumulated value since the 1980s for the Netherlands20. The Dutch net-worth of foreign assets (NFA) is included
in this figure as well. Comparing the accumulated value of current account surpluses and the net foreign asset position is disturbing. The value of the NFA-position is far less than the accumulated value of current account surpluses. Moreover its development is volatile and seems uncorrelated with the accumulation of past current account surpluses. While the Netherlands have had a yearly surplus since the 1980s and the NFA-position was positive at that time, the NFA-position even fell to € –113 billion in 2002. Recent statistics by the Dutch central bank report that the NFA-balance has turned into a surplus again. Currently it amounts to €105 billion (DNB, 2009). This exemplifies the volatility of the NFA-account. All in all, the development of the Dutch NFA-position is surprising. Standard macroeconomic accounting suggests that the identity “current account” ≡ “net savings” should hold. One would, thus, expect that at least the path of NFA and the accumulated current account balance to be somewhat related. Somehow the value of the Dutch accumulated current account is absorbed into a “black hole”. Is there any explanation?
20 Note that actual accumulated value should be higher as the balance over the years previous to 1980 is positive as well (DNB, 2009).
25 Figure 1.13 Net Foreign Assets and Current Account compared (billions of euros)
Source: DNB (2008a)
The phenomenon was first identified for the Netherlands by Kusters (1997) and recently been brought to attention by Boonstra (2003, 2004). This “black-hole phenomena” is not unique for the Netherlands . For instance, Hausmann and Sturznegger (2007) have also successfully established this same discrepancy between NFA-position and accumulated current account balances (termed “dark matter”) empirically for a set of 103 countries.
Mellens and Noordman (2009) provide an overview of the debate, including the arguments made Hausmann and Sturznegger. First, the value of the NFA-position is not a simply a mirror of past current account surpluses. Instead, the NFA is determined by four factors: fluctuations in exchange rates, asset prices, different other economic factors (like deductions on goodwill, poor cash flows and degrees of risk) and statistical errors. Mellens and Noordman find that exchange rate and asset price effects are the source for 80% of revaluations. It is difficult to assess the economic implications of the “black hole”. Fluctuations in exchange rates and asset prices do not have a direct and immediate effect on the real economy as long as they are not related to factors of production. That is, as long as the asset positions are not materialised and turned into cash to finance consumption or investment they are of little relevance. However, these fluctuations are of concern when the NFA is used to pay for imports, as will be assumed in this thesis. In this case changes in the asset position affect relative prices, as they determine the volume of imports. Mellens and Noordman make a number of useful comments. First, in the long run price effects can be set at zero and thus disregarded. Second, it seems that in comparison to foreign investors, Dutch investments abroad have lower risk-profiles
-150 0 150 300 450 600 1986 1990 1994 1998 2002 2006 Year Billion euros NFA
Current Account Balance
Accumulated Current Account Balance
26 and yield lower returns. That lower risks yield lower returns, is common financial knowledge. Nevertheless, in general, expectations about future returns on Dutch foreign assets are too optimistic and should thus be scaled down. Finally, since the value of NFA follows from a complex configuration of determinants, the NFA should not be targeted as a policy variable. In sum, Mellens and
Noordman’s study suggest that we should not be too concerned about short-run fluctuations when analysing the long-run, yet the return on past savings might be lower than generally considered.
With these considerations in mind, it is possible to look at the projections on the NFA. Both Van Ewijk et al. (2006) and Bettendorf and Knaap (2003) have modelled a business-cycle neutral path for the NFA-postion, suitable for long run analysis. These estimates are included in table 1.5 as well. In Bettendorf and Knaap’s projection, clearly, dissaving takes off in midst of the 2030, during the heydays of ageing (around 2035-2040). NFA as a percentage of GDP falls considerably. Van Ewijk
et al. (2006) index the development of NFA by taking 2006 as a base year.21 Again, there is an initial
positive balance. Dissaving, in contrast to Bettendorf and Knaap, is not as substantial. The figures in table 1.5, as taken from both studies, illustrate that future paths of the NFA are difficult to assess and can be different. Still, both studies give an alike trend. NFA will continue to increase (a lot) until the 2030s. In the second part of the 2030s, the value of the NFA will start to fall as the current account balance turns into deficit. Nevertheless the sizes of projected NFA seem large enough to cover expenses on imports.
21 Though the index is not as informative as an actual figure or a share of GDP-value, most relevant is its development. For this reason, publication of the index suffices.
