This is an appendix for ‘Not in Narrow Seas: New Zealand History from an Economic Perspective’, a book I am writing. It is published here so that people can access the technical material. The data is available on request.
Keywords: Macroeconomics & Money; Political Economy & History; Statistics;
Over the last sixty years we have got into the habit of thinking about the course of the economy as ongoing exponential growth with ever rising standards of living – small hiccups aside. The rate of that underlying growth – the secular trend – might vary with some periods slower than others, some faster, while imposed over any secular trend is a (relatively regular) business cycle which might even have periods of brief retardation. But its amplitude is small relative to the long term growth trend. Of course in the distant past there were periods – centuries and millennia – of economic stagnation (or perhaps very, very slow growth) but some time ago – among the nineteenth century North Atlantic economies – the process of ongoing exponential growth began as a continuing part of economic experience.
Many of the generalisations in the preceding paragraph are contestable – even wrong. In particular while without question output per head has risen substantially in the last 150 years (by around ten times), the pattern – for New Zealand anyway – has not been a simple story of exponential growth. There have been long periods of stagnation (‘long recessions’ or ‘depressions’), in all covering more than a third of the period, and over half of the first eighty years.
While the specific interest is to describe this historical evolution of the New Zealand economy, the experience is also of contemporary interest. Forecasters implicitly or explicitly, assume a growth trend fitting a cycle around it. Getting that (probably changing in the long term) secular trend right is crucial; the bias, arising from recent history, towards a rising trend rather than the possibility of a longish period of stagnation could mislead the short term forecasting.
Some Tangential Issues
There are some issues which may be triggered by the data in this chapter even though they are, to some extent, tangential to its central concern. They are dealt with immediately to enable a focus on the course of the New Zealand market economy up to 1939. The series’ definitions and problems are dealt with in the appendix.
Welfare and Output
Is Gross Domestic Product (or some related measure of output or income) per person a measure of welfare of a nation? While an unqualified ‘yes’ is often attributed to the economics profession it has never been true that they all uncritically equated national output with welfare. Simon Kuznets, creator of the statistical base from which GDP derives, wrote in his original report in 1934 ‘the welfare of a nation can scarcely be inferred from a measurement of national income’, while John Kenneth Galbraith wrote an elegant chapter decrying per capita GDP as a measure of welfare in his Affluent Society published in 1958. Less well-trained politicians, journalists and business people like to talk about GDP as if it is a measure of welfare claiming that all economists agree with them – even when they dont.
Why, if economists have known for last 75 years that GDP was inadequate as a measure of welfare have they not constructed a better measure? The short answer is that they have tried, and they have not been able to. (Non-economists have tried too, but without understanding the complexities or subtleties of the exercise.)
A committee led by Joseph Stiglitz and Amartya Sen, two contemporary giants of the profession, is considering the matter again. Their preliminary report concluded there is no single measure of economic welfare, but there may be a number of indicators (thereby providing a critique of the non-economist’s attempts to replace GDP with another single indicator). Had a unique single measure of a nation’s welfare been possible, economists would have developed it shortly after 1934. One must never assume that the best economists are as stupid as their critics.
The difficulties of using output (or income) measures as an indicator of welfare have been reinforced by recent research which suggests that the correlation between per capita output (or real income) and self-reported happiness is tenuous, at least among the more affluent countries. At the very least, doubling of the indicator does not double happiness – New Zealanders are certainly not ten times happier today than they were in 1860 – it may not increase it much at all. This is a burgeoning research area – currently fraught with paradoxes and without any accepted comprehensive conclusion. But unquestionably it underlines how wrong it is to equate output/income with welfare.
(The research suggests significant gains in happiness from an output increase in less affluent countries. It may not be coincidental that the foundations of utilitarian theory were constructed in the early nineteenth century, when average Western Europeans had a lower per capita income than those in most of today’s poor countries; perhaps the assumption that increased consumption increased happiness may have been closer to the truth then.)
The reason that this chapter – and indeed the book – is interested in GDP is because it measures output, despite changes in output not reflecting changes in people’s welfare, nor changes in the way people live. Arguably the changing the composition of out put (including new products) and the changing methods of production (including new processes, some of which enable the increasing leisure) which changes the lives people lead, rather than the rise in their material consumption.
Economic Growth and Sustainability.
The view that economies depend on unlimited economic growth cannot be true since economies have stagnated for literally millennia; the view that the study of economics depends upon unlimited economic growth cannot be true either since many giants of the economics profession – including Thomas Malthus, David Ricardo, Karl Marx, John Maynard Keynes and Joseph Schumpeter – were stagnationists who expected economic growth to come to an end. Keynes wrote of the ‘euthanasia of the rentier’ because, as capital accumulated, the return on capital would fall, until there would be no incentive to invest, and economic growth would stop; many other economists thought so too.
The difficulty with this stagnationist expectation is that per capita incomes in the rich world quadrupled in the 180 odd years between Ricardo and Schumpeter. By the 1950s, as the data became available, it became evident that a theory of economic growth dominated by pure capital accumulation was inconsistent with the facts.
The stagnationist models tended to underplay the importance of technical change which enables the same capital and labour to produce more (or different sorts) of products. That was the insight of Robert Solow’s famous 1957 paper. About this time the belief of ongoing exponential growth became more common, eventually to dominate the way we think about the economies The implication is that as long as technical progress continues there will be rising output in the long run. When it ceases, the stagnationist expectation will become relevant again. (This is a long term prognosis. In the short term there may a business cycle which involves temporary stagnation or contraction.)
Nevertheless, it can be argued that the current economic system is dependent upon economic growth since the true profit rate is close to the growth rate in the long run; so no growth means no profits. Since it is difficult to conceive of the existing system functioning on the basis of zero profits – a variation of the Keynesian ‘euthanasia of the rentier’ thesis – when growth exhausts itself the nature of the economic system would change.
That this chapter is exploring exponential growth – or the lack of it – does not mean that it assumes that such growth is inevitable. Concerns about unlimited growth are irrelevant except for the future. The issue here is the past growth record.
There are two further major defects of GDP. First it may not include the output of subsistence producers, especially farmers, which is consumed by the producer and possibly swapped without being commercially traded in the market. Many settlers in the first three quarters of the nineteenth century were subsistence farmers and most Maori remained so well into the twentieth.
Domestic Non-market Production
Second, since domestic production (production in the household which is not paid for) generally has no market value (that is, it is not traded in the market), it does not usually appear in GDP. That does not mean it is unimportant or lacks social value. Its omission is simply a consequence of the way GDP is defined (and the difficulties of otherwise measuring it, for despite trying economists have not been able to find a robust way to incorporate domestic activity into a wider output measure). Perhaps domestic production might be treated as the largest subsistence sector; it persists as such, although probably on a reduced relative scale scale compared with the past, to this day.
This omission has generated much anxiety because it seems to devalue domestic activity – much of which is women’s work. In economic history, there is a different problem. The domestic (in the sense of housework et al) sector has also experienced major productivity improvements over the years. How large has not been unmeasured, but presumably they are not exactly the same as in the market sector, while the relative productivity growth between the sectors probably varied; it is likely there were periods in which domestic productivity rose faster than market activity.
This effect cannot be ignored, because of the interface between two sectors. Some of the contributions to domestic productivity – better houses – appear as capital goods in the market sector, others, such as domestic appliances like refrigerators and vacuum cleaners, are consumer goods. Much of the domestic productivity came from outsourcing as food and clothing which are now purchased but were produced in the home in the past; child care is a service example. The domestic productivity improvements increased household leisure but they also released labour into the market sector – first domestic servants, later as (married) women worked for remuneration.
Given the lack of material – other than anecdote – there is little we can do here, except be aware of the limitations. We return to the issue more fully in a later chapter.
In a modern economy population growth chugs along at around 1 percent p.a. Variations affect the economy in part through the demand for facilities and the resulting stimulation of the building and construction industry. In the medium to long term, shifts – such as the aging of the population – may matter. But population change in the nineteenth century – even if we ignore the decline of the Maori, and subsume it into the total – was dramatic in proportional terms.
A good approximation is that between 1861/62 and 1938/39 New Zealand’s population increased by 19,000 each year (although the dominant cause of the increase shifted from migration, to births and then an offsetting mortality). That meant an increase of around 1 percent on the end 1938/39 population of 1.6m but over 10 percent on the beginning population of 180,000 in 1861/62.
Rapid population growth from immigration generates an air of prosperity which is not evident in the per capita GDP figures. The migrants may well feel incomes are higher than their origin economies (assuming unemployment was no worse). Their arrival creates opportunities for entrepreneurs, while increasing the population – a doubling occurred in the decade in the 1860s – may lead to gains from economies of scale. There may even be, as a result, a general lift in incomes of the established population and from the additional rents that property owners would expect to receive.
We shall return to the population issue below; it is probably not so important after the ‘Vogel’ period other than the loss of manpower during the war, which is covered in the next chapter.
The Quantitative Evidence: 1861/62-1938/39
As explained in the appendix there are two independent series – the Rankin series and the Greasley-Oxley series – of per capita volume GDP which start in the middle of the nineteenth century. They run from 1861/2, 1938/39. (All the GDP data discussed in this chapter are ‘volume’ or ‘real’, that is adjusted for changes in the overall price level, and per capita. )
Over the 77 years to 1938/39 the Rankin series shows a trend increase of 0.8 percent p.a. while the Greasley-Oxley series shows an increase of 0.5 percent p.a. Both figures may be misleadingly high because they do not allow for the shift from subsistence production to market production. They are both low by the standards of the post -Second World War economy.
The difference of 0.3 percent p.a. between the two series in their mean rate of growth may reflect measurement error, although a 0.3 percentage annual points compounds to a 25 percent difference over 77 years. Some of the difference would arise if service sectors volume rose faster than commodity sector ones, for the Greasley-Oxley series is based only on the commodity sectors (agriculture, mining and manufacturing) and largely omits the service sector. It may well underestimate total GDP volume growth. That should not – by itself – affect the medium term swings around the long term trend. In the following we shall be inclined toward treating the Rankin long run growth rate – derived from spot estimates – as closer to reality.
Figure 1 shows the raw data from 1861/2 to 1938/39. It is very jerky, so in order to reduce measurement noise in the series and to override the three year business cycle is also shown. This smoothing however, blunts the points at which there is a change in the underlying trend. (We know very little about the business cycles in this period, although we would also be surprised if both series had the same cyclical turning points.)
Are the long term deviations from the secular trend consistent with the qualitative historical narratives? Can we refine the narrative with the data?
The 1860s to about 1875
Both series show above trend output in the 1860s, although the patterns are rather different. The Rankin Series is one of steady growth to about 1875, with no evidence of a mining boom, perhaps because many of the transactions would have been carried out with (sometimes foreign) coins which are not included in the estimate of the money stock, on which the series is based.
The Greasley-Oxley series shows a boom peaking in 1869, and a marked contraction to about 1875. One suspects that had it included the building and construction sector, the peak would have been earlier, and the contraction sharper The G-O series conforms more closely to the historical narrative. When he launched his borrowing and development program in 1870 Vogel would have been puzzled by the story the Rankin series tells, but a G-O series with building and construction was closer to Vogel’s account of the times.
Of course Vogel did not have the GDP figures. His indicators were what was happening in business as the alluvial gold and war economies ran down. The resulting slowing of immigration (and hence population growth) would also have depressed the economy, or at least reduced business opportunities (following our earlier discussion on the impact of immigration).
In summary, the available GDP series are supportive, and such discrepancies as there are between them and the traditional narrative can be better explained by their limitations rather than because of major errors in the narrative.
From the mid-1870s the Rankin Series per capita GDP remains much the same until about 1895, although real income was slightly higher in the early part of the period. Thus the data is consistent with the notion of a long recession.
The Greasley-Oxley Series shows some growth from the mid 1870s to the mid 1890s. The level of output is below the long term tend but roughly matching it. Again the lack of the building and construction sector may be distorting the pattern. The public works program initiated by Vogel hardly appears in the sectors the series covers. Without this sector, the series does not really tell a story of a ‘Long Depression’ – perhaps the best it tells is of a ‘growth recession’, a period of modest but below capacity growth. (The moving average does not return to its 1869 peak until 1891.)
The data series supports the narrative describing a long period in the 1870s, 1880s and 1890s in which economic growth was, at best, subdued. Allowing for its limitations it suggests a long recession (or the ‘Long Depression’) although the data is arguably closer to the Sutch account of the economy being depressed or slow growing from the late 1860s. 
1895 to 1910
Both series show an acceleration of the growth rate from sometime towards the end of the nineteenth century. Between 1895 and 1910 the Rankin Series shows a per capita increase of about a third (an annual growth rate of about 2 percent, more than double its long run rate). The Greasley-Oxley series shows a 19 percent increase between 1895 and 1902, a 2.8 percent annual increase, over five times it average secular rate.
Despite the differences in the peak both series show a substantial acceleration of the growth rate from the mid 1890s, but the acceleration peters out for both in the following decade.
Most historical narratives imply rising prosperity in the early years of the Liberal Government, although there is a tendency in the narratives to locate the upturn earlier (perhaps because they see it as a consequence of government policies). Nevertheless one gets from them a sense that in the latter stages of the first decade of the twentieth century the Liberal government was struggling. This is usually attributed to Ward taking over from Seddon, but it may be the economy began stagnating and Seddon was lucky (in this narrow sense) to get out while his reputation was still intact.
From the end of the growth acceleration (1909/10 in the case of the Rankin Series, 1901/2 in the case of the Greasley-Oxley Series) output per head once more stagnates.
In the case of the Rankin Series, the stagnation is until the Great Depression which commenced in 1929, when there was a sharp contraction.
The Greasley-Oxley Series story is a little more complicated. Basically the output per capita track is level from 1901/2 to 1916/7, and then there is a fall of about 8 percent in the four years to 1920/1 after which the series broadly flattens off again until 1928/9 – perhaps there is some weakening at the end of the period. It is also a more volatile series than the Rankin one. If we moderate the track by allowing for a faster growing service sector then it is flatter but still weaker in the 1920s compared to the 1910s.
The Lineham Series shows a volume growth rate of 1 percent p.a. from 1917/18 to 1928/29, which seems more prosperous than the historical narrative. The Chapple Series is too short (from 1926/7) to identify any trends but seems to report a mild contraction from its beginning through to the Great Depression.
There are a number of complications in the the First World War and interwar period but the basic conclusion is that the twenty odd years were, at best. very subdued in terms of per capita output and probably stagnant. Contemporary accounts are broadly consistent with that; their memory was largely swept away by the Great Depression.
The Great Depression: 1928/29-1934/35
The seven year moving average method does not work well for this period because while there was a sharp contraction and while the depression was deep it was also short – only seven years. That is very evident in all the raw series, and will be discussed in the relevant chapter.
All the output per head series show a sharp rise after 1934/35. This was not a depression recovery; the series tend to show per capita output at above the 1928/29 level by 1935/36.
Later we shall consider the evidence that the strong growth path continued into the 1940s, Given that this period covers about half of the boom we leave it here.
While the statistical series present a number of challenges of interpretation they are broadly consistent with the narrative. What is really important is they demonstrate that for over half of the eighty years after 1860 – from at least 1875 to 1895 and from (probably) 1910 to 1935 – output per capita was largely stagnant or falling.
NOTES TO DEFINING AND MEASURING OUTPUT
Definitions of GDP
This Chapter uses GDP because it is the most readily available indicator of output, not because it is a perfect measure. To simplify, nominal Gross Domestic Product is the market value of the goods and services produced in a particular area. One of the influences on that change is price changes. Real (or volume) GDP involves using the same set of market prices (often the prices for a given year) on each period’s output, in order to eliminate the effects of price changes – of inflation and deflation. However it is not a perfect measure because it can be influenced by the choice of the particular set of market prices, especially important in a small open economy which can experience major changes in the relative price of exports and imports – the terms of trade.
It is usual to scale GDP by measuring it relative to population. Over long periods the composition of population – including the proportion of adults in the workforce – changes so per capita (real) GDP does not give a reliable indication of productivity change.
Even in narrowest economics terms per capita GDP is not the best aggregate measuring welfare through time. A better aggregate would be National Income measured in constant expenditure prices rather than production prices. A major difference between the two are that depreciation of capital which is deducted from GDP (and is probably not important) and that expenditure and production prices differ by the terms of trade.
Another adjustment is instead of measuring income generate in the country – domestic  – to measure the income generated by the residents of the country – national. The largest source of this difference for the New Zealand economy is that some of the domestic income is earned by investors who live off shore; without a corresponding offset of New Zealand investors obtaining a similar magnitude from their offshore investments.
The various deductions give National Income. The choice of GDP in this narrative is largely because it is the measure used in the national discourse. But attention will be drawn to any difference where the distinction matters (other than that GDP tends to be higher).
Measures of GDP
The main text is based on a series which is synthesised from various series. Here is an account of them before the synthesis. (There are various simplifications in this account which does not purport to be an authoritative review of the series, and which omits some caveats.)
As will become apparent, the indexes used to measure GDP are of varying quality so it is necessary to say something about their origins. As a general rule, the further back a series reports the less reliable it is.
Statistics New Zealand: SNA Nominal and Volume Series 1977/8 – to present
Statistics New Zealand (SNZ) has produced GDP related series back to (March year) 1931/32, albeit on different conceptual bases (and subject to different degrees of accuracy). Undoubtedly the best are the SNA (international System of National Accounts) estimates which go back to 1977/8. They are calculated on both the expenditure and income sides of the national accounts which provides a cross-check, and are available both nominally and in volume terms. The latter estimates are based upon sectoral output series either deflating nominal measures or using a direct volume indicator (e.g. kilos of wool produced).
Statistics New Zealand: ONA Nominal and Volume Series 1954/5 – 1977/8
The earlier comprehensive SNZ series, labelled ONA (Old National Accounts), was calculated primarily on the nominal income side of the national accounts with the expenditure side components calculated separately except for the private consumption which was a residual.
They were originally measured as GNP, that is the output generated by New Zealand nationals, rather than the output generated in New Zealand, but can be readily converted to GDP if the latter information is available.
Volume GDP was calculated by various methods on the expenditure side (but not on the output side) back to 1954/55.
A particular issue is that this series and the later (SNA) one do not overlap properly at 1977/78, because of an inventory measurement problem. It does not matter much in the long term, but does matter for medium term assessments and led to misleading understandings in the 1980s.
Statistics New Zealand: ONA Nominal Series 1938/9 – 1954/5
Constructed in a similar manner to the later (post-1954/55) with some gaps during the war (which can be interpolated). However there is no official volume series. The conversion was done by the In Stormy Seas approach described below.
Statistics New Zealand Nominal Series: 1931/32 – 1937/38
This may not be an ‘official series’, but it is reported in some New Zealand Official Yearbooks and so is more ‘official’ than the unofficial ones described below. It is only nominal, and constructed as a measure of national income (i.e. on the income side), with no expenditure side estimates at all. It can be scaled to GDP, and can be converted to a volume series by the GDP deflator method as for the ONA from 1938/39 to 19454/55.
Chapple Series: 1925/6-1937/8. 
Simon Chapple took the SNZ series from 1931/2 to 1938/39, added an earlier estimate for 1925/26 and interpolated the intervening years. 
Lineham Nominal Series: 1917/18-1938/39. 
Brent Lineham constructed a nominal series based on sectoral outputs – which formally is the correct way to measure GDP. He did not construct a volume series.
The Stormy Seas Volume Series: 1917/18-1954/55.
In my In Stormy Seas: The Post-War New Zealand Economy I used the ONA series, the SNZ 1931/2 to 1938/39 series, and the Lineham series (slightly improved at the beginning with a better farm output series) – deflated by a GDP price index to provide a volume GDP series for the period. (The GDP deflator was a weighted average of the available long term price indexes.)
G-O: Greasley-Oxley Series: 1860/61 – 1938/39.
David Greasley and Les Oxley constructed a volume output series based upon recorded exports and domestic sales. As such, the G-O measurement corresponds most closely to that used in official (SNA) measures of real GDP series. However it suffers from the following deficiencies:
1. It only the farm, manufacturing and mining sectors, excluding the others which made up half of the economy in the inter-war period. Use of the series as a direct indicator of the economy as a whole requires the assumption that the measured sector to unmeasured sector ratio has been reasonably constant. However, the other sectors grew about 2 percent p.a. faster in nominal terms than the three measured sectors in the interwar period. This may, however, reflect rising relative prices rather than faster relative volume increases.
(Insofar as the other sectors have a different sectoral pattern – the building and construction sector in particular tends to be more volatile than other sectors – their exclusion may mean the G-O series does not capture the business cycle well.)
2. It probably does not cover the subsistence sector well. This is not so much a weakness of the series per se, but of all series covering the nineteenth century unless specific attention is paid to the subsistence sector. None of the New Zealand series do.
3. The method assumes that there are no major changes to inventories. Normally this would affect the business cycle, but in addition there were major inventory buildups of wool, meat and dairy products during the First World War (towards the end sometimes amounting to over a year of production), followed by an unwinding in the early 1920s. (There were also build up of wool on wealthier farms during the Great Depression.)
Unfortunately the late 1910s and early 1920s period is important in statistical terms because it is a lap period between series. So I replaced the farm output series for the period 1915 to 1925 with an output series constructed by Todd Simpson.
(It was not necessary to do this for the Great Depression wool stocks , which were smaller, the turn round was quicker, and it is not in a lap period.)
The GO series does not align on the available spot nineteenth century estimates.
Rankin Series: 1858/59 to 1938/39.
Keith Rankin constructed a series which goes back slightly earlier than the Gresham-Oxley series, using contemporary estimates with a money-multiplier method to interpolate between them. The method had been earlier used in New Zealand by Gary Hawke. (below), It is a synthetic (indirect) approach since it assumes a relationship between GDP and another aggregate, in this case the stock of money.
He describes his method thus: ‘The quality of my data [GNP, not GDP]is based first and foremost on the quality of the contemporary estimates around which they are based. Secondly it is based on the quality of the Australian non-monetary estimates. The interpolations were based on velocity estimates drawn from patterns in Australian macro data, and not on Hawke’s assumption that the dates of the Australian business cycle matched the dates of the NZ cycle. Also crucial to my methodology was the verification of differences in the Australian and NZ cycles with the trans-Tasman migration data.’ 
I have scaled up the GNP series on the assumption of a constant ratio to GDP. Presumably the estimates do not cover the subsistence sector, since it does not have a lot of to do with money. An additional complication is that the series is calculated as a nominal aggregate, and deflated by the consumer price index which is not a good indicator of the GDP deflator.
Hawke Series; 1870-1918. 
The Hawke series, which follows a similar Brazilian approach, used money multipliers without calibrating on the available estimates. It is only available as a nominal series. Hawke says the ‘technique produces a stop-gap rather than a substitute’ and estimated derived ‘from records of factor rewards or products’. 
Madison Series: 1-2008. 
The Maddison series is a comprehensive data base of world GDP and population through the Common Era. Intermittent estimates for New Zealand go back to 1000CE (sic) but are annual
from 1870. The period to 1938/39 are based on the Rankin series.
The strength of the Maddison data base is that it allows comparisons with other countries. However, as the 1000 AD New Zealand GDP shows, it may not be up-to-date. Latest understandings suggest there were no humans in the economy at that time.
Because the 1870 to 1939 figures are based on the Rankin Series they are not treated separately in the main text.
There is an official population data base going back to 1840. In each period it is probably a lot more accurate than the best available GDP estimates, although care needs to be taken for the inclusion or exclusion of the Maori population, the estimates for which are less reliable.
The Labour Force
In order to calculate average labour productivity it is necessary to have a labour force series. Unfortunately the ideal is near impossible for the early years, and consistency requires the same definition for later years.
The best that could be done was to construct a series based on those who reported their membership of the labour force in census returns and interpolate where annual estimates are not available. In the early years the census did not ask Maori for their workforce status. These were extrapolated by assuming similar participation rates by age to the non-Maori labour force although Maori were more likely to be in the subsistence sector. .
It was not possible to adjust for unemployment since there are no census data before 1896; in any case the notion may be less meaningful in the nineteenth century. Because unemployment is more volatile between censuses than labour force participation it is not sensible to interpolate it or employment numbers. Nor is it possible to adjust for hours worked until the late 1940s. The productivity series is therefore indicative and interesting but not even as reliable as the GDP per capita series. So the resulting ‘productivity’ series is not reported in this chapter.
This Appendix has benefited from detailed comments by Keith Rankin.
 S. Kuznets (1934) ‘National Income, 1929-1932′ 73rd US Congress, 2d session, Senate document no. 124.
 J. K. Galbraith (1958) The Affluent Society
 Report of the Commission on the Measurement of Economic Performance Et Social Progress (2009) http://www.stiglitz-sen-fitoussi.fr/documents/rapport_anglais.pdf
 R. Lanyard (2005) Happiness: Lessons From A New Science
 J. M. Keynes (1936) The General Theory of of Employment, Interest, and Money, Book VI, Chapter 24, Concluding Notes
 R. Solow (1957) ‘Technical Change and the Aggregate Production Function’. Review of Economics and Statistics 39 (3): 312–320.
 Confusingly economics uses the term ‘domestic’ in two quite distinct ways. In Gross Domestic Product it refers to the boundaries of a jurisdiction, so that New Zealand GDP refers to the production within the boundaries of New Zealand, which is different from Gross National Product which refers to the production by New Zealand citizens (nationals) which may occur offshore, but excludes production by non-citizens inside New Zealand, most notably the returns on their investments here. But ‘domestic’ production also refers to non-market economic activity in the household (which is typically not included in GDP). To reduce the confusion, the term ‘domestic’ in the second sense will have the word ‘household’ close to it.
 The Greasley-Oxley series measures the production gross, which means that the outputs of the service sector which are inputs to the commodity sectors are included in their gross output.
 I added 0.3 percent p.a. to the G-O growth rate, bringing the long term trends of the two series to much the same. However the adjusted G-O series has the late 19th century growth boom from the mid 1880s rather than the Rankin series mid 1890s, and peaks at the beginning of the 1900s rather than closer to 1910.
 An ambiguity arises since the exact period of the data estimate may be end of the defined year, its middle, or the average level over the year. Without quarterly data, business cycle identification can be haphazard – with it, the exercise merely becomes difficult..
 W. B. Sutch (1957) Long Depression.
 See footnote 7.
 B. Easton (1992) The 1977/78 Downturn in the New Zealand Economy http://www.eastonbh.ac.nz/?p=76
 S. Chapple (1994) ‘How Great was the Depression in New Zealand? Neglected Estimates of Interwar Aggregate Income’, New Zealand Economic Papers, p.195-203. See also K. Rankin (1994) comment p.205-209.
 For interpolation he used nominal factory value added, which may not be the best series for this purpose. To give a level comparable to the three other series (which are based on 1938/9 = 1000), it is based on their average in 1937/8, there being no 1938/39 one.
 B. T. Lineham (1968) “ New Zealand’s Gross Domestic Product 1918/38′, New Zealand Economic Papers, Vol 2, No 2, !968, p. 15-26.
 B. H. Easton (1997) In Stormy Seas: The Post-War New Zealand Economy, Appendix A5.
 B.H. Easton(1990) A GDP Deflator Series for New Zealand: 1913/4-1976/7, Massey Economic Papers, B9004.
 D. Greasley & L. Oxley (2008) Reinventing New Zealand: Institutions Output Patents 1870-1939, Working Paper 15/2008, Department of Economics, University of Canterbury.
 Lineham op cit, p.16
 Note that the G-O outputs are measured gross, so any inputs from other sectors into the measured ones are included. It could be argued that the difference in long term growth rates between the Rankin and G-O series reflects this omission. The gap is consistent with what the 2 percent p.a. might suggest.
 T. E. Simpson (1991) The Origins of Statutory Producer Control Legislation in New Zealand Agriculture: 1914-1925, MCA Thesis, VUW, Appendix 1.
 K. Rankin (1992) “New Zealand’s Gross National Product: 1859-1939” Review of Income and Wealth, 38, No 1 pp.49-69.
 K. Rankin (2010) per com. 14, October 2010.
 G. R. Hawke (1975) “Income Estimation from Monetary Data: Further Explorations”, Review of Income and Wealth, 21, pp.301-307.
 Op. cit. p.306.