The NIER's forecasts cover the next two to three years. Medium-term projections are also produced for subsequent years, the most recent extending to 2020. This paper explains some of the principles underlying these medium-term projections and how they are performed.
Every forecast is surrounded by some uncertainty, and the farther ahead we look, the greater the uncertainty. It is particularly hard to tell where the Swedish economy will be in its cycle three years or more down the road. Cyclical fluctuations arise when the economy is exposed to unexpected shocks, which affect it for several years thereafter. Gradually the influence of historical shocks subsides, and unexpected new shocks, by definition, are not foreseen.
The repercussions, however, may continue for a long time, and the effect of historical shocks may still be present in the initial years of the medium term. But the impact on these years is normally less than on the years covered by short-term forecasts. In other words, forecast cyclical fluctuations diminish the farther ahead we look.
Forecasts for the more distant future are generally based almost entirely on long-term trends in factors like productivity and number of hours worked. Attempts to determine in some other way where Sweden will be in the business cycle in seven years, for example, will probably result in greater forecasting error.
On account of the substantial uncertainty, the medium-term projections are used primarily as a reference path for alternative scenarios, i.e. as a basis for discussions of alternative assumptions. Here the calculations must be consistent, the projection for a particular economic variable must be reasonable, given that the rest of the economy develops as projected.
If there are no unexpected shocks, the economy will gradually follow its long-term trend. But the long-term trend does not necessarily coincide with the historical trend; changes in demographics, for instance, may affect factors like the size and composition of the labour force. With the aging population of Sweden, for example, the country's labour force will probably grow more slowly in the period ahead than its average rate of change over the last ten years.
This is an instance of an expected shock, which is considered in the medium-term projections.
Support in calculating the adjustment of the economy towards its long-term trend is provided by the NIER's dynamic macroeconomic model KIMOD (see NIER Working Paper No. 100). Among other things, the model takes into account the imperfect information of the players in the economy and the consequent inertia of prices and wages that delays restoration of equilibrium in the market economy.
In the long run the development of the economy is determined by its supply side, i. e. the supply of labour, level of education, level of technology and availability of investment capital. The total number of hours worked in the economy is calculated in this case from demographic forecasts of the working-age population and estimates of the equilibrium unemployment rate, the proportion of the population outside the labour force and average hours worked. (The NIER's method is described in Chapter 3.2 "Hours, Capital and Technology — What Means Most? — the Development of Productivity Analyzed with the Use of Growth Accounting," Appendix to the Long Term Planning Commission Study (Långtidsutredningen), February 2008.)
GDP is calculated as the total number of hours worked multiplied by productivity, i. e. the value of the goods and services generated by each hour worked. Productivity is projected at the industry level and aggregated to the national level with consideration given to expected changes in the composition of the business sector and in the relative size of the public sector in the period ahead (see "Productivity and Wages Through 2015 (Produktivitet och löner till 2015)", Special Study No 6, May 2005, or the above-mentioned Appendix to the Long Term Planning Commission Study (Långtidsutredningen).
Sweden currently has a large trade surplus, net exports are positive, resulting in annual current-account surpluses in relation to other countries. But the current-account surpluses are not expected to last in the long run. If they did, there would be an endless increase in the real net wealth of Swedes abroad — an unreasonable supposition. For the medium-term it is assumed that the composition of GDP will gradually shift, with a decrease in the balance of trade (net exports), in proportion to GDP, and an increase in domestic use of GDP.
One important reason for this assumption is that the population is aging. An aging population means both that the current income of older people decreases so that they use up their savings, and that demand for domestically produced welfare services like health and elderly care may increase in proportion to GDP. Such a development is consistent with a decrease in the GDP share of exporting industries.
If net exports are lower (in proportion to GDP), the total of gross investment, general government consumption and household consumption must provide a larger share of GDP. The calculation of general government consumption in the medium term is based on the assumption that employment in the general government sector will adjust to the demographic trend so that personnel density (per inhabitant of a particular age and gender) is held constant in public-sector operations in order to maintain an unchanged living standard. General government consumption is also assumed to adjust to the surplus target for general government finances; see below.
The net lending of the general government sector, i. e., the difference between the revenue and expenditure of the sector, is estimated to be slightly above 1 percent of GDP in 2020. In practice this outcome can be achieved in a number of ways, but in the medium-term projections it is assumed that expenditure on general government consumption and on transfers to households will be adjusted so that the surplus target is met. FIMO, the NIER's fiscal-policy model, is used in performing the calculations.
As long as the applicable rules remain the same, general government consumption expenditure will depend chiefly on changes in the size and composition of the population. The level is corrected somewhat in the medium-term projections in order to reflect in part the adjustment to the surplus target. This step will have repercussions on the composition of GDP, including exports and household consumption. General government transfer payments to households (sickness benefits, unemployment benefits, etc.) are projected partly on the basis of demographics and employment, partly according to the rate of increase in prices and/or wages. The projections of transfer payments are thus more mechanical for the medium term than in the short-term forecast, for example, by not taking behavioural effects into account to the same extent. Moreover, a transfer to households is included "for calculation purposes" so that the surplus target is met.
The revenue of the general government sector consists largely of taxes. In the medium-term it is assumed that the tax bases will develop at the same rate as the respective real-economic variables like household consumption or total earnings, whereas ("implicit") tax rates are assumed to remain unchanged after the final year of the short-term forecast. An exception is made for capital taxes in relation to GDP, which are assumed to adjust to the historical average after a few years.
The assumption that general government finances will be in balance in the long run is consistent with the assumptions that GDP will approach its long-term trend and that general equilibrium will prevail. Weakening general government finances will tend to give households a disposable income consistent with the consumption forecast and domestic saving (net lending) consistent with smaller surpluses in the current account.
The picture of the real-economy generated in the medium-term projections is consistent with the repo rate forecast published in The Swedish Economy. Here allowance is also made for the likelihood that expansionary fiscal policy will increase inflationary pressures and thus necessitate tighter monetary policy. In the long run, inflation will approach the Riksbank's target of 2 percent per year. Wages and demand for labour and capital will thus be affected by the repo rate so that real unit labour costs are consistent in the long run with a given internationally set rate of return (read about the NIER's conceptual framework for the development of labour costs in the Wage Formation in Sweden 2006). The exchange rate is also determined in the projections; the decrease in net exports is accompanied by appreciation of the Swedish krona.