Productivity growth in Norway 1948-2008.

• Author: Hagelund, Kare

1. Introduction

   Productivity developments are reported regularly in Statistics Norway's Economic Survey (Statistics Norway, 2009). Norges Bank has also discussed productivity growth previously in boxes in its Inflation Report (Norges Bank, 2006) and Monetary Policy Report (Norges Bank, 2007). Halvorsen (2006) provides an overview of productivity growth in the period 1981-2003. The present article analyses productivity growth over a longer period and attempts to date breaks in underlying productivity growth. As it has to be assumed that technological advances in Norway are associated with innovations abroad, there is also a comparison with productivity growth in other countries.

   A country's productivity growth determines growth in standards of living over time. With productivity growth of 3 per cent, GDP can double in 25 years without needing any increase in the number of hours worked; with growth of 1 per cent, this will take 70 years.

   The crucial importance of productivity growth for economic growth over time can also be illustrated as follows: In the period from 1948 to 2008, GDP for mainland Norway increased by an average of 3.3 per cent per year. Productivity growth contributed 3.1 percentage points of this, whereas the number of hours worked contributed, on average, just 0.2 percentage point. GDP growth since the war has therefore been driven primarily by growth in labour productivity.

   Productivity is closely related to potential output--in other words, the maximum level of output compatible

   with stable price and cost inflation. Estimates of productivity growth are therefore an important part of the basis for monetary policy. The high global inflation of the 1970s and early 1980s was probably a result of overly high estimates of growth in potential output, which illustrates the importance of quickly capturing changes in underlying productivity growth (see Orphanides et al., 2000). In theory, and over time, productivity growth will also affect the long-term equilibrium real interest rate (see, for example, Bernhardsen, 2005).2 Changes in this interest rate are important for how expansionary or contractionary a given key policy rate will be for the economy. In isolation, higher productivity growth implies a higher long-term equilibrium rate, as the required rate of return in the economy increases.

   Productivity growth is affected by factors such as the population's education, the size and quality of fixed capital, levels of research and development, infrastructure, the production and use of ICT, opportunities to trade with other countries, levels of state and foreign ownership, restructuring in the private and public sector, and how well legal and financial institutions are functioning. The design of the welfare state, the level of wage differentials, the population's age composition, labour force participation and workers' training and skills development, the degree of competition and macroeconomic conditions will also impact on productivity growth. In the light of all these factors that can influence productivity, it is not surprising that there is no generally accepted theory for what determines productivity or for how it should be measured.

   Section 2 below provides a brief overview of different measures of productivity and the measurement problems associated with different concepts of productivity. Section 3 analyses productivity growth in Norway, while Section 4 compares it with that in other countries. The article concludes with some thoughts about how productivity growth may be affected by the financial crisis.

2. Different measures of productivity

   For an enterprise, productivity expresses how efficiently it uses inputs such as labour and capital equipment to produce goods and services. If an enterprise produces more goods and services with the same inputs, or the same quantity of goods and services with fewer inputs, productivity has increased.

   Productivity is typically defined as a ratio between a volume measure of output and a volume measure of inputs. Which measures are used will depend partly on the purpose of the analysis. Productivity can be defined relative to a single input (such as labour) or multiple inputs (such as labour and capital). As volume measures of output and inputs are associated with considerable uncertainty, calculations of productivity are subject to substantial measurement problems.

   Labour productivity

   Labour productivity is used in many analyses. This is a relatively simple but important measure of productivity. It is also closely related to growth in real incomes. One key tenet of economic theory is that the rewards of labour are determined by labour productivity. Labour productivity depends on factors such as organisation, logistics, incentives, use of technology, qualifications and the quantity and quality of fixed capital. It is also influenced by how intensively capital and labour are used.

   The simplest measure of labour productivity is output per worker. Output per worker increases if more is produced per hour worked, but also if more hours are worked. During an economic upswing, hours worked per worker normally rise. If productivity is measured as output per worker, one might mistakenly conclude that underlying productivity (and so potential growth) is changing over the business cycle even if output per hour worked is constant. If productivity is measured as output per worker, we might therefore mistakenly register an upswing in productivity and possibly overestimate growth in potential output. In that case, we might subsequently be surprised by higher-than-expected inflation.

   Output per hour is therefore a better measure of productivity. The advantage of this measure is that it takes account of the fact that the number of hours worked per person can vary. One disadvantage is that the number of hours worked is harder to measure than the number of workers. In international comparisons, it is also normally easier to use the number of workers than hours worked, because statistics for the latter are not as standardised and are normally less accessible. As the number of hours worked per worker has changed considerably over time and also varies a great deal over the business cycle, however, we will be using output per hour in this article.

   Labour productivity can change as a result of technological advances or an increase in capital equipment (or another input). One weakness of partial productivity measures such as labour productivity may be that allowance is made for increased inputs of labour but not for increased inputs of capital. This means that the measure does not necessarily reflect technological advances alone.

   Total factor productivity

   It is usually said that an increase in output which is not attributable to an increase in the use of inputs such as labour and capital reflects a change in total factor productivity (TFP). TFP can, in principle, be calculated by taking growth in output and adjusting it for the contribution from specific inputs. Besides new, improved technology, an increase in TFP may be due to better logistics, more efficient utilisation of premises, or other changes in the organisation of production.

   In practice, TFP cannot be observed and has to be calculated. There are various ways of doing this. In the calculations in this article, we impose an explicit production structure. We assume that the production structure in the mainland economy can be described by the following simple relationship between value added (3), labour and capital services: (4)

   (1) [Y.sub.t] = [A.sub.t] [K.sup.(1-[alpha]).sub.t] [L.sup.[alpha].sub.t]

   Measurement problems

   Y = Value added

   K = Capital services

   L = Labour, measured as number of hours worked

   [alpha] = Wages paid as a proportion of value added

   A = Change in output not attributable to primary inputs (TFP)

   As we are using value added, we can ignore factors of production other than capital and labour. We also assume that TFP is not linked to any of the factors of production but is Hicks-neutral. Increased TFP will not then affect the desired amount of capital per worker and will, in principle, represent "pure" (and cost-free) technological progress.

   When decomposing growth in labour productivity into contribution from growth in capital intensity (the increase in capital services per hour worked) and contribution from growth in TFP ([g.sub.tfp]), we use equation (2), which is derived simply from equation (1). [g.sub.Y], [g.sub.L] and [g.sub.K] are growth in value added, hours worked and capital services respectively.

   (2) ([g.sub.y]-[g.sub.L]) } Growth in labour productivity = (1-[alpha])([g.sub.K]--[g.sub.L]) } Growth in contribution from capital intensity + [g.sub.tfp]

   Measuring fixed capital's contribution to output is associated with significant challenges. In principle, it is the volume of the services that this capital contributes to output that is to be included. In the calculations performed here, we assume that the volume of capital services is proportional with the capital stock as estimated in the national accounts.

   The quality of labour evolves over time. We measure labour input as hours worked regardless of quality. Changes in...