From In Stormy Seas, p. 200-208.
Keywords: Growth & Innovation;
Here we use the term (fixed) capital to mean physical items such as land improvements, buildings, plant and equipment. It excludes land, and it excludes inventories (physical stocks). Sometimes physical capital is thought of as `stored labour’ to distinguish it from the ongoing labour which accompanies its use. The term `capital’ is also used for financial assets. In principle these are matched by physical capital after the netting out of liabilities. Financial capital is not a concern of this chapter.
The capital measure used here is valued at cost less depreciation, so called `net value’. This does not necessarily reflect the productivity of the capital. For instance, the government built a second oil fired power station at Marsden Point (Marsden B) which is unlikely ever to be used. Its commercial value is near zero. But it was valued in net capital at around $250 million (in current prices), even though the resources and effort implied by that total was almost entirely wasted. Thus the measures of capital used here allow us to focus on the extent to which the capital investment has been productive or not.
In March 1950 aggregate capital, measured in 1981/82 prices was $51.8 billion dollars, about 4.6 times the GDP (also measured in 1981/82 prices).(2) By March 1990 the capital stock was $149.0 billion, or 4.4 times GDP (Philpott 1991a). The annual growth of capital stock was 2.7 per cent, fractionally below the 2.8 per cent p.a. of volume GDP. In fact the rate of growth varied over the period. In broad terms capital stock grew about 3.4 per cent a year, between March 1950 and March 1977. In the following 13 years it averages 1.6 per cent a year, with a fade at the end of the 1980s.
Figure 6.4 [not included – see page 201] shows the ratio of total capital to GDP. In the short term a rise in the ratio indicates a recession in New Zealand, a fall represents a boom. But there are also longer term trends. Between 1950 and 1975 the average capital to output ratio was falling, indicating that the economy was growing faster than its stock of capital. After that period the capital to output ratio increases, although by 1990 the ratio has not recovered to its 1950 level. That means. In effect, that New Zealand was using its capital stock increasingly more efficiently over the post war era. Allowing for the more depressed state of the economy in the late 1980s, the improvement in capital use appears to apply over the whole period, except for the early 1980s. This rise is the result of the Major Projects (or `think big’) which involved large extremely capital intensive industries.
Interpreting the average capital to output ratio is not easy. The higher it is, the more capital intensive is production. Some means of production are highly capital intensive, and yet very valuable to the economy. Living in a dwelling is a capital intensive activity, for there is little (paid) labour used in that activity (repairs and maintenance excepted). There is also little labour in running some great capital enterprises, such as the Maui platform. On the other hand a high capital to output ratio may reflect poor productivity, as in the case of the useless Marsden B power station.
The Sectoral Pattern of Capital
An indication of the changing pattern of sectoral capital-output intensity is shown in table 6.6, where the sectors have been classified into high, medium and low. The picture consistent with the aggregate figures is of a slight fall in capital intensity. There are four general points to be made about the sectoral pattern.
CAPITAL OUTPUT RATIO
|Owner Occupied Dwellings||36.4||35.2||-3.4|
|Electricity, Gas, Water||25.6||12.7||– 50.6|
|Food & Beverages||3.4||2.3||– 31.3|
|Wood Products||2.8||1.5||– 44.3|
|Pulp & Paper Products||3.2||2.0||– 36.5|
|Finance, Real Estate, &
|Textiles & Apparel||1.7||1.1||– 32.1|
|Non Metal Manufactures||1.2||2.0||65.5|
|Wholesale & Retail Trade,
Hotels, & Restaurants
Source: Philpott & Nana (1985)
1. The three high intensity sectors are dwellings, electricity, and farming. The latter two experienced major declines. In the case of electricity it probably reflects the rise of thermal power stations, which have a lower capital intensity than hydro ones. The decline in farming in part reflects major productivity gains, and in part the falling profitability of farming discussed in Chapter 5II.
2. The manufacturing sector in New Zealand is not especially capital intensive. Some of the least capital intensive activities, and hence more labour intensive, are fabricated metals (includes car assembly) and textiles.
3. There are a few sectors which experience a major change in capital intensity. In three cases the industry structure was transformed:
– in mining it is the Maui Platform in the seas off Taranaki in the late 1970s (and to a lesser extent the onshore Kapuni field earlier);
– in chemicals it is the oil refinery at Marsden Point in mid 1960s, and the various gas based activities in Taranaki in the 1980s (the ammonia urea plant, the methanol plant, and the synthetic petrol plant);
– in base metals it includes the aluminium smelter at Te Wai Point and the steel mill at Glenbrook, both instigated in the 1960s and extended in the early 1980s.
The fourth sector which increased its capital intensity was business and financial services. (This excludes financial assets of the sector.) The expansion was about equal in both plant and equipment (including computers) and in buildings.
4. There is a high ratio for the public service. This is mainly because of the inclusion of roading and suchlike for which there is no attributed output, thus raising the sector capital to output ratio.
International Comparisons of Capital to Output Ratios
How does New Zealand’s capital intensity compare with other countries? Only a few have official capital series. They are shown in Table 6.7. There is a wide variation between the 10 OECD countries for which there is data, probably because of measurement differences. The New Zealand estimates include roading and a more extensive definition of improvements to land, whereas others tend to be less comprehensive about these items. Some countries seem to be very mean in their estimates of non-market (i.e. government services) capital. To reduce the composition effects (and the definitional problems which are more acute in some sectors) the tabulation provides the average capital to output ratio with some sectors’ capital omitted.
The overall impression is that in terms of fixed capital, without inventories, New Zealand has an above average capital to output ratio. The most comparable ratio is the last column: for total GDP to manufacturing, other industries, and services capital (without the housing, government non-market, or agriculture sectors.) It puts New Zealand at about 2.0, whereas the ratio for the OECD 10 range from 1.0 for the United States to 1.9 for Greece, with a mean of 1.4.
It is not difficult to provide anecdotes why the New Zealand capital output ratio may be high.(3) It is frequently claimed that
– marginal land broken in was uneconomic;
– the design and implementation of some capital investment was inefficient;(4)
– much of the energy investment in the 1970s appears to be in unutilized at the time;
– freezing works had to spend substantial amounts in the 1970s to upgrade their facilities to meet overseas hygiene standards, and so on.
INTERNATIONAL CAPITAL OUTPUT RATIOS: 1982
Sources: New Zealand – see text; Others: OECD (1989)
Notes: * 1980, ** 1982/83.
It is also plausible to argue that inflation distorted investment decisions, encouraging capital investment (in buildings for instance) for capital gain. However there is not a lot of systematic evaluation of such effects.(5)
It is common to measure the efficiency of `factor inputs’ (labour and capital) by using a ratio of (constant price) output (usually measured by value added) to labour (measured in workers, or hours worked), called (average) labour productivity. This ignores the contribution of other factors such as capital stock, usually because there are no measures of it. However before reporting an assessment of the efficiency of both factors, it is useful to report briefly one detailed study on labour productivity.
This New Zealand study was responding to some overseas findings which had observed a slow down in the growth of labour productivity in the 1970s. It found a parallel decline at the economy wide level and the sector level here, also from the mid 1970s. However Phillipa Marks found about half of the observed decline could be attributed to a cyclical downturn, for productivity tends to grow more slowly when the economy stagnates and production capacity becomes under-utilised. Among the explanations which were explored but proved not to have had a significant effect were inter-sectoral shifts in employment, changes in the labour force composition, and the capital to labour ratio. Measurement errors were mentioned as a possible unquantifiable source of the unexplained slow down, and macroeconomic explanations were also considered as a potentially fruitful area for further investigation, but this was not explored (Marks 1983).
Labour productivity – the ratio between total output and labour input – is an easy enough concept to understand. The rest of this section is going to use a somewhat more complicated idea of total factor productivity (TFP). The following example will illustrate the notion. Suppose both the labour input and the capital input to an economy (or sector) grew 2 per cent, and the output grew 3 per cent. Then we might attribute the difference between the two (i.e. 3-2 = 1 per cent) to improvements in the productivity of the factors. This is called TFP.
The annual (or whatever period) increases in TFP measured this way may be linked together in a `TFP index’, which Robert Solow labelled in a seminal paper `technical change’. His term is somewhat misleading as he noted.
“I am using the phrase `technical change’ as a shorthand expression for any kind of shift in the production function. Thus slowdowns, speedups, improvements in the labour force and all sorts of things will appear as `technical change’.” (1957:312 original’s italics).
In fact the TFP index is an arithmetic residual, not a variable with explanatory power in its own right. Some pressure groups – especially those in the educational and technological sectors – have been keen to claim their activities promote the index. But Tom Balogh and Paul Streeten shrewdly described it as the `coefficient of ignorance’ (1961), that is the part of the growth process we cannot empirically explain.
Adrian Orr took up the challenge laid down by Phillipa Marks, using total factor productivity in a sectoral and cyclical analysis. When comparing the two studies note that the later study had additional data (i.e. 1960/61 to 1986/87 instead of 1961/62 to 1980/81), almost doubling that available after the mid 1970s. Orr could not find a distinctive structural break in the mid 1970s. Rather he argued that `a temporary breakdown in cyclical behaviour of productivity occurred’ (1989:63) during the 1974/75 to 1977/78 period, appearing to disrupt the trend.
I came to a similar conclusion in parallel work(6) looking at the 1961/62 to 1987/88 period at an aggregate level adjusting for hours worked (rather than the headcount workforce), and for capacity and lag effects (Easton 1989a). There was evidence of a slight productivity decline over the period, but it was not statistically significant. Splitting the period into two halves (at 1975/76) I found the total factor productivity index rose 11.8 per cent in the first half and 11.5 per cent in the second.
Orr also provides sectoral TFP estimates, at a greater detail. (Table 6.8) The international data is dealt with later.)
SECTORAL TFP GROWTH (1960/61-1985/86)
Per cent p.a
|SECTOR||New Zealand||OECD Average|
|Mining and Quarrying||4.7||-1.3|
|Electricity, Gas, Water||5.0|
|Trade, Restaurants, Hotels||-0.4||0.8|
|Transport & Storage||1.8||(2.4|
|Finance & Business Services||-1.0||-1.0|
Source: Orr (1989:77); OECD Average: Englander and Mittelstädt (1990:22) – 13 countries.
After adjusting for the business cycle and for hours work, I found an average growth rate of 1.0 per cent a year. Allowing then for these effects, production relative to (used) labour and capital inputs increased by 1.0 per cent a year, 10.4 per cent a decade, or 48.9 per cent between 1950 and 1990.(8) Roughly, then, it could be said that output per worker hour has increased 50 per cent by factors other than additional capital and labour.
In total, 60 per cent of measured economic growth over the period (1961/62 to 1987/88) could be accounted for by increased aggregate hours worked with the same amount of capital per hour, 11 per cent from additional capital per hour work (capital deepening), and the remaining 29 per cent from TFP (`technological change’). Focusing on the contributions to additional output per worker hour (i.e. labour productivity), we find just over a quarter (27 per cent) of that was due to capital deepening and almost three quarters (73 per cent) to technological change.
The figures assume GDP volume growth is accurately measured. However if as Chapter 1 observed, the growth rate could be underestimated by about .3 per cent a year. If this is correct then the annual TFP growth would be 1.3 per cent a year.(9)
A further adjustment arises from land, for the above studies only looked at fixed capital. In effect some of the labour and capital had to be used on the land as a substitute because there was no additional land of equivalent fertility, an illustration of the law of diminishing returns. Calculating the effect of this is problematic, but a heuristic calculation suggests land supply effects would add not more than .1 per cent a year to the TFP growth rate, raising it to 1.4 per cent a year.
How does New Zealand’s TFP growth rate compare to that of other OECD countries?
International Comparisons of TFP Change
The best international TFP data is for the market (or `business’) sector only. Market TFP increases about .4 per cent a year more than total TFP.(10) That would put my adjusted figure near 1.8 per cent p.a. (or 1.4 per cent p.a. ignoring measurement errors and land).
The OECD has published cross-country comparisons of TFP change, which include New Zealand (Englander & Mittelstädt 1990, OECD 1991). Unfortunately this involves the suspect assumption of treating each country the same, ignoring their individual cyclical experiences. The studies found that there was a lower rate of TFP growth across all OECD countries, in recent years, which was attributed to `structural’ factors: the end of post-war construction, the reduced scope of catchup [discussed below], less rapid expansion of international trade, slowdown of technological advances, changes in the composition of the labour force, and perhaps [sic] increased government regulations’ (1990:48).
The TFP growth rates for each OECD country for the period from 1960 to 1990 are shown in table 6.9. The OECD’s estimated average annual TFP growth for the New Zealand business sector is .4 per cent, in comparison to 1.6 per cent for the OECD average. The figure is surely too low, and wildly different from the New Zealand based estimates of business TFP growth of about 1.4 to 1.8 per cent p.a.(11) Using adjusted New Zealand based estimates we might conclude that the New Zealand TFP growth was about the same as the OECD average, or perhaps fractionally below. Note however, that the OECD (weighted by GDP) average is dominated by the poor showing of the United States, with the (country) median being 2.0 per cent p.a. New Zealand is a definitely a little below most OECD countries.
Table 6.8 also gave the OECD sectoral estimates for 1970 to 1983 with Orr’s figures for 1960/61 to 1985/86. The basic impression is that the New Zealand service sector TFPs are lower, further evidence that there is a measurement error in our GDP growth rate. In almost every other sector the New Zealand figure is internationally respectable, and in some cases (especially agriculture) it is impressively above average.
OECD TFP GROWTH RATES (1960-1990)
Business Sector Only
Per cent p.a.
Source: OECD (1991:136)
The Determinants of Total Factor Productivity Growth
Ideally we would like to reduce the `coefficient of ignorance’ by splitting out other effects which contributed to economic growth. However as discussed earlier, there is no adequate data base on changes in the quality of the labour force. Another item which overseas studies have attempted to allow for is the gains from research and development, but again this has not been possible for New Zealand.(12)
We have come to three general conclusions about New Zealand TFP.
– Its growth level seems to be similar to (perhaps a little below) the OECD average, especially if there are measurement problems with the New Zealand service sector.
– Insofar as New Zealand’s TFP is slightly below the OECD median, it may be due to the relatively higher population growth. (That effect was estimated to be about .17 percent per annum.)
– Unlike the rest of the OECD there was not a decline in the New Zealand TFP growth rate in the late 1970s and 1980s, compared to the 1960s and early 1970s. This means its TFP growth was probably below the OECD average in the earlier period, and higher than the later period.
Why did not New Zealand show the same secular decline in its TFP growth rate? One answer might be that the New Zealand TFP is not influenced by the rest of the OECD, although a major thrust of this study is that New Zealand is very influenced by the rest of the world.
An alternative explanation centres on the `catch-up’ theory, which argues that it is much harder to pioneer a technology than it is to imitate it. Thus countries which are more technologically backward tend to have higher TFP growths than the most advanced countries, as they take on the technologies developed by others. Thus, it is argued, there is a tendency for poorer countries to catch up with rich ones.(13)
What does the catch-up effect mean for New Zealand? In the 1950s and the early part of the 1960s the country had an above average GDP per capita, and so its TFP growth should have been below the OECD average, as other countries caught it up. However by the late 1970s, after the great post 1966 decline, New Zealand was now well below the OECD average, and the catch-up effect should have been positive, which would give it an above average TFP growth – which is what we observe.
That, of course, does not reverse the three generalisations about TFP growth. In its own terms the second appears to reject the hypothesis that long term TFP growth has been higher in periods of higher GDP volume growth (Philpott 1985). The evidence is that it is independent of the growth rate, since the latter varied over the 30 odd years, but TFP did not.
The Supply-side and the Poor Growth Performance
What we have seen is that compared to the actual GDP growth performance, labour supply (adjusted for productivity) grew faster. Thus it is not evident that labour shortages inhibited the New Zealand economic growth. Though while international comparisons are not really possible, the fragmentary data suggests that the quality of the New Zealand labour force is not outstanding. The evidence does point to New Zealand using its capital less efficiently than the other OECD countries, although there may well have been gains in efficiency over the period. All this would explain the poor level of GDP per capita, not the decline.
If we use Total Factor Productivity as a measure of how effectively the factors of production were used, we again find that New Zealand was not markedly out of line with the rest of the OECD, although one acknowledges that such comparisons are fraught with difficulties. But even if the New Zealand TFP growth was below average, there remains a major problem. It was broadly constant over the period. Thus it does not explain the variations in the long term growth rate.
Thus we can rule out the supply-side as a major explanation for the poor post-war economic performance, although high population growth in the earlier stage probably had a small depressing effect.
1. Bryan Philpott has made numerous contributions to this study. But I am especially grateful for his tenacity and energy which has produced the capital series used here, and for his generously permitting me to use his data.
2. Because this ratio is measured in constant (1982/3) prices it is not comparable with the ratio shown in Figure 6.4 which is in current prices. The current price ratio is higher in the 1950s because the price of capital goods tends to rise more slowly than the price of all goods and services.
3. Adjusting for relative prices between capital and GDP using the 1990 purchasing power parity study only slightly reduces the New Zealand ratio. But it remains well above the average.
4. `Gold plated’ irrigation headworks (paid by the state rather than the farmer) were alleged in an investigation in which I was involved. Marsden B power station is another example.
5. The claim of a high capital to output ratio is supported by the evidence of the investment to GDP ratio, which is average among OECD countries (ESC 1984:50-2). It follows that the productivity of our investment is lower than average, but this is an algebraic conclusion deriving from the poor growth rate and the average ratio. It tells us nothing about causality. If there was anything else inhibiting the growth rate – for instance shortages of foreign exchange limiting expansion – then we might expect an average investor to have a poor return on their investment.
6. Unfortunately neither I nor Orr realised the other was working in the same field until after publication, and did not compare notes.
7. Orr gives TFP change estimates for individual sectors. Unfortunately the employment figures on which they are based are wrong. The manufacturing figure is calculated by weighting (1973/74 valued added) Orr’s subsector figures, and are probably robust to any non-linearities. I made the same mistake in Easton (1987), and the manufacturing subsectoral analysis in that paper should be ignored.
8. Assuming the rate applied throughout the whole period.
9. In which just over three quarters (78 percent) of the labour productivity gains attributed to `technology changes’.
10. Orr (1989:77) tabulates average annual TFP growth in the non-service sector was -.9 percent, while the figure for the economy as a whole was .9 percent. Suppose the performance of the non-market sector (which is in the service sector) was the same for services as a whole and for market sector services. Under this assumption market TFP would have increased by 1.3 percent a year, or .4 percent more than the total economy.
11. A tabulation in Englander & Mittelstädt (1990:14-15) indicate that an hours work adjustment would raise annual TFP growth by about .2 percent points. This may explain a small part of the divergence.
Bryan Philpott thinks that the OECD capital stock series may be misleading, but because the details of the construction of the OECD data are not published it is not possible to check this.
The inclusion of the 1985 to 1990 period almost certainly depressed the average because of the increased under-utilization of capacity.
12. A comparative study in the late 1980s found New Zealand R&D was lower as a proportion of GDP than for a number of other comparable small OECD countries (Edwards 1992). It was especially lower in the private sector, and for manufacturing and social research. However R&D on agriculture was about double that of average.
13. This theory has to be empirically tested, for one can equally think of just as plausible theories which give the innovators a permanent and increasing headstart. In fact both Dowrick and Nguyen (1989) and Englander and Mittelstädt (1990) find a catch up effect among OECD countries, where one might expect the technology transfer to be easier.
Balogh, T. & P. Streeten (1963) `The Coefficient of Ignorance’, Bulletin of the Oxford University Institute of Economic and Statistics, May 1963, p.99-107.
Dowrick, S. & D.T.Nguyen (1989) `OECD Comparative Growth 1950-1985: Catch-up and Convergence’ American Economic Review, Vol 79 No 5, December 1989, p.1010-1030.
Easton, B.H. (1989) Aggregate Technical Change: 1961/2-1987/8, Hodge Fellowship Report, Wellington.
Easton B.H. (1993) `Review of “Private Pensions in New Zealand: Can They Avert the Crisis?”‘ NZEP, 27(2), p.255-259.
Economic Summit Conference (1984) A Briefing on the New Zealand Economy, Wellington.
Economist (1990) Book of Vital World Statistics, Hutchinson, London.
Englander, A.S. & A. Mittelstädt (1990?) `Total Factor Productivity: Macroeconomic and Structural Aspects of the Slowdown’, OECD Reviews.
Edwards, F. (1992) Research and Development Spending: A Comparison between New Zealand and Other OECD Countries, Report 5, Ministry of Research, Science, and Technology, Wellington.
Gould, J. (1982) The Rake’s Progress, Hodder & Stoughton, Auckland.
Marks, P.A. (1983) `The Slowdown in Labour Productivity Growth in the 1970s’, in B.H. Easton (ed) Studies in the Labour Market, NZIER Research Paper No 29, Wellington.
OECD (1989) Flows and Stocks of Fixed Capital: 1962-1987, OECD Department of Economics and Statistics.
OECD (1991) OECD Outlook, OECD, Paris.
Orr A. (1989) Productivity Trends and Cycles in New Zealand, NZIER Research Monograph 48.
Philpott, B.P. (1985) Economic Research and Economic Policy, NZIER Discussion Paper 29, NZIER Wellington.
Philpott, B.P. (1990) `Immigration and the Economy’ in Migration and New Zealand Society: Proceedings of the Stout Centre Annual Conference, 1989, Stout Research Centre, p.128-137.
Philpott, B.P. (1991a) Provisional Estimates of New Zealand Real Capital Stock by SNA Production Groups: 1950-1990, RPEP Internal; Paper 233, Wellington.
Philpott, B.P. (1991b) `Economic Growth in New Zealand: Models and Experience’, in L. Evans, J. Poot, & N. Quigley (ed) Long-run Perspectives on the New Zealand Economy Philpott, B.P. & G. Nana (1985) Database of Sectoral Output, Labour and Capital Employed 1959/60 to 1983/4, RPEP Internal Paper No 184, Wellington.
Solow, R.M. (1957) `Technical Change and the Aggregate Production Function’, , August 1957, 39.