A Research Gathering: Viewing New Zealand’s Social, Economic and Political History Through the Eyes of the CPI; 15 July, 2010
Keywords: Political Economy & History; Statistics;
My task today is to convey that the Consumer Price Index (CPI) occurs in an economic, political and social context. That does not mean that the statistic lacks authority, nor that its compiler – today, Statistics New Zealand – lacks independence. But, as we shall see, the history and development of the index – any social statistic – has to be seen in a broader setting.
That two or three apparently independent events which occurred in the latter end of the first decade of the 1900s, led to the establishment of the CPI, suggests that there was an underlying driver. Almost certainly that was the end of the Long Depression, about a decade earlier.
There were not good measures of output in those days and periods of stagnation were identified by falling prices. British prices had been falling from about 1875 and bottomed around 1895, indicative that the New Zealand Long Depression was part of an international phenomenon. By the end of the nineteenth century, prices had begun to rise. A similar pattern is evident in the available New Zealand series, although the fall had begun at least a decade earlier. Perhaps that reflected the ending of supply shortages from boom caused by the alluvial gold in the south and the war in the north; possibly, too, international transport cost were falling. The implication is that the British price level was an important influence on the New Zealand price level; there was after all, a full monetary union.
Presumably these were the factors which led James Hight, lecturer in economics and history at Canterbury University College from 1901 and professor from 1909, to suggest to James McIlraith to investigate the course of prices for his LLD (there were no PhDs in those days). The resulting thesis, The Course of Prices in New Zealand: An Inquiry into the Nature and Causes of the Variations in the Standard of Value in New Zealand, was published by the Government Printer in 1911 because, as Hight explains in the introduction, ‘there is no University Press’ in New Zealand. (The Introduction is interestingly defensive about the use of indexes at all: ‘The index number is now regarded as an indispensable instrument in an inquiry into questions affected by changes in the power of money’, pointing out that they have been ‘available for some time … in England and America’.)
McIlraith averaged the wholesale prices he gathered to get his index; he does not seem to have had access to expenditure weights from, say, a representative household budget survey. His interest seems to be the relation of the overall price level to gold – and marriage.
Back in 1840 the New Zealand Government had been instructed by the Colonial Secretary in London to collect data on prices – presumably reflecting the Victoria appetite for ‘facts’, for the resulting statistics seem to have been published but were not greatly used until McIlraith’s construction of a variety of indexes from 1861 to 1910, later extending the series to 1913.  McIlraith is interested in general price levels, and does not specifically estimate a consumer price index, although there is some material there towards the construction of one.
In his 1911 introduction Hight mentions the ‘recent rise in the cost of living’. This was the second thread. Prices seemed to be rising. We have already noted this was an international phenomenon, but the usual local suspects were blamed including, unsurprisingly, wages; especially blamed was the Industrial Conciliation and Arbitration Act established in 1894. It is true that about that year prices bottomed but the coincidence of two events does not prove causation, nor that New Zealand’s IC&A Act caused English prices to start rising too.
There seems to have been a sufficient public outcry for the Minister of Internal Affairs – at the time his department included the Registrar General’s Office which was responsible for official statistics – to offer a public rebuttal in 1908. John Findlay pointed out that between 1895 and 1907 an index of wages rose 23 (actually 23.7) percent and the price of provisions ‘on the bare necessaries of life’ rose 22.0 percent (actually 22.5 percent), concluding that ‘wages and prices for necessary foods had advanced at nearly equal rates in thirteen years.’  Despite the price index covering only food this may be the first occasion that a fledging (and incomplete) consumer price index appeared officially. 
The third event, possibly related to the second, but also with independent roots was initiated by the ‘Harvester’ decision in which the Australian Court of Arbitration attempted to fix a wage specifically designed to guarantee to a worker a certain standard of living.  The New Zealand court was aware of the decision and its Judge even said he thought it was the Court’s duty to establish a minimum living wage for the lowest paid workers. However, except in 1908 when it wrote ‘We think that anything less than 7s per day is not a living wage where the worker has to maintain a wife and children’, the Court itself said nothing significant about a living wage prior to the Great War.
Perhaps one other event contributed to the development of the official index. In 1910 the Census and Statistics Act created the Office of the Government Statistician. Within a couple of years the 1912 Royal Commission on the Cost of Living was established chaired by Edward Tregear, previously Secretary of Labour. Hight was one of its members, and McIlraith, described as its principal witness, was closely and intelligently cross-examined by his mentor. It reported within the year. The 13 questions posed to the Commission suggest there was a wider remit – perhaps even a review of the economy – but whoever drew them up was obviously not an economist.
The Official Series Gets Underway
At the head of the Commission’s recommendations was that official statistics needed to be greatly improved. Apparently the government responded with additional funding and a ‘mini-boom’ in statistics followed. Indeed in 1914 the office may have been at its peak of brilliance with Douglas Copland working on price statistics and John Condliffe working on external statistics; both were economics students of Hight and both went on to distinguished international careers. It is Copland who begins the official development of what was to become the Consumer Price Index, although he was limited by available data and covered only food, fuel, lighting and housing or about 60 percent of household expenditure, omitting apparel and services, among other things.
It is said that ‘experience is what you get just after you need it’. It is more usual to get statistics long after they are needed. But in the case of the CPI the Copland and successors’ data series soon proved relevant. Its origins had reflected a change in the world view; after decades of falling in the late nineteenth century, consumer prices began rising towards it end. While the rises may have been a calamity in terms of what was then considered the declining norm, they were small; according to Margaret Arnold-Galt’s consumer price series from their nadir in 1895 they rose by 28 percent by 1914 – about 1.3 percent p.a. There was a barely noticeable acceleration towards the end of the period – 1.9 percent p.a. in the last five years. 
But in the next five years to 1919 the rise was 8.4 percent p.a., an aggregate increase of 50 percent. Marvellously the official statisticians already had underway an – albeit incomplete – framework to monitor the change.
The inflation was, of course, the consequences of financing the Great War, here and offshore, through monetarisation. It disturbed traditional relativities. Much of that story belongs elsewhere, but one of the pressures was on wages (complicated at first by awards only being able to be updated every three years). The Court of Arbitration had begun to place some reliance upon the fragmentary price base as early as 1912.  Following a change in legislation the Court announced in April 1919 that in addition to basic wages ‘the Court will grant to workers a bonus by way of compensation in the cost-of-living and this bonus will be varied from time to time according to the rise or fall in the cost-of-living as ascertained by the Dominion [sic].’ 
The 1920s and 1930s
This led to a famous incident in 1920 – explored in detail by George Wood – in which the Court awarded a 9s a week cost-of-living bonus and then shortly after replaced it with a bonus 3s a week (almost 2d an hour when an unskilled wage was 1s 3½d an hour) because of an ‘erroneous’ misunderstanding of the data presented by the acting Government Statistician.  The essence of the problem – it appeared again in the double-digit inflation era of the late 1970s, but in political squabbles – was that increases can be measured point on point or period average on period average. Under low inflation the difference may not matter much, but when there is high and variable inflation, it does.
Focussing on the kerfuffle obscures a creeping structural change. The Court was no longer primarily concerned with a fair wage but was increasingly concerned with the impact of price increases, despite the insertion of the expression of ‘fair standard of living’ briefly in the IC&A Act in 1921-2. The focus then was some notion of ‘real wages’ (‘real consumption wages’).
Wage fixing was under considerable pressure following the disruption from the inflation during the Great War and the stagnation of the 1920s. Meanwhile the statistics office was improving – as best it could – the quality of its measurement of consumer prices. The implication – often overlooked – is that earlier parts of any continuous Consumer Price Index series are far inferior to later ones in terms of quality, coverage, comprehensiveness and accuracy. Those who use the series over the long term as if it is statistically consistent do so at some peril. An indicator is the expression ‘Consumer Price Index’ was introduced in 1948; previously it had been called the Retail Price Index although it included items not sold by retailers.
(An additional complication is a deeply technical one – what exactly is the underlying conceptual basis of a Consumer Price Index? Unravelling the issue belongs to a paper yet to be written.)
However there is a practical political economy issue, especially around the treatment of housing, which Arnold-Galt observes has been revised every time the overall regiment has been reviewed.  The difference may not be trivial for while the prices of most items in the regimen follow the median and, in any case, their weighting is too small for a different path to matter much, housing outlays are both a large part of consumer spending, and tend to rise markedly faster than any median. Hence the treatment of housing can markedly affect the course of the index. For instance following the 1955 revision, the owner occupied housing component included the going price of a house (which, not incidentally, is an asset price). Had that approach been used in the CPI of the first decade of the 2000s, the CPI might well have been increasing at 0.5 or more percent p.a. than was reported. (At which point the counterfactual becomes unclear, because the Reserve Bank might have been forced to maintain a more restrictive monetary stance in order to keep within the target zone specified in the Policy Targets Agreement, unless it was changed too.)
Illustrative of the ambiguity arising from the lack of rigorous underpinning, has been a tendency to use interchangeably the expression ‘cost-of-living’ with CPI, a practice which persisted long after the Government Statistician had eschewed the former term as meaningless.  Ironically the first statutory reference to consumer prices in the IC&A Act is in a 1936 amendment introducing General Wage Orders, in which the Court was required ‘to have regard to the general and financial conditions then affecting trade and industry in New Zealand, the cost of living, and any fluctuations in the cost of living since the last order.’ Previously the Court had taken it into consideration, but now there was a statutory direction. 
The Second World War
The legislation was suspended in December 1942, when the government introduced a comprehensive plan for economic stabilisation. (Lurking in the memory was surely the inflation during the Great War.) Key to the scheme was a Wartimes Price Index, with the requirement that if it rose by more than 5 percent there would be a corresponding increase in wages. except the first order was to be made if they rose by more than 2½ percent. The WPI consisted of 238 items (including, for the first time, fresh fruit and vegetables). The Department, which collected the prices, continued to compile the Retail Price Index but it was not published until after the war.
The temptation to manipulate the Wartime Price Index was irresistible. It rose less than 2.5 percent between March 1943 and September 1947, that is under the threshold which would have triggered the first general wage increase. Meanwhile the Retail Price Index rose 7.4 percent – three times as muc (and given that some of the wage costs would have been passed on, enough to trigger the second wage increase.)
There were at least three ways by which the manipulation could occur. The Price Tribunals controlling prices could resist increases in the designated items, allowing cost recovery in the non-designated ones. Items could be subsidised; by 1945-6 subsidies on essential clothing and foodstuffs amounted to £3.0m or over 1 percent of consuming spending and there were additional subsidies on the inputs of shipping transport, coal production and distribution and wheat, amounting to another £2.5m.
If these options failed, there were more direct responses. Bernard Ashwin, a chairman of the Stabilisation Commission, later recounted : “I remember on one occasion the statistician’s report recorded that prices had in fact increased by more than [the threshold]. I immediately called the Government Statistician and it turned out he had included the price of onions – which couldn’t be brought anyway – in his price calculations. I told him the country’s economy was not going to be upset by the price of onions and so we withdrew the item and got below the [threshold] again. The country was saved!” 
George Wood noted that the more general practice was ‘when a commodity went off a the market temporarily – and this happened quite often – the price [in the schedule] was held until supplies became available again … This meant that the more frequently shortages occurred the easier it was to maintain the index at a stable level’. 
Certainly compared to the Great War, price rises in the Second World War were modest; between 1939 and 1946 the increase was 21.5 percent, a fraction less than 2.9 percent p.a. Open inflation became a more pressing issue after the war ended.
The Wartime Price Index included tobacco, but not beer, which was also omitted from the Retail Price Index. During the review of the regimen in 1949, Wood, now Government Statistician, was approached by F. P. (Jack) Walsh on behalf of Prime Minister Peter Fraser, who had reacted with horror when an earlier minute mentioned the possibility of including beer. (Fraser consumed vast quantities of weak tea and buttered bread.) ‘So beer was out.’ The committee, which Walsh had ‘stitched’ up, explained that only Britain included beer prices in its index, while four other English speaking countries did not. 
Wood was greatly hurt by this political intervention, the only occasion of which I know. Fraser was still thinking in terms of the index being a measure of the cost of living, which would contain only ‘approved’ items. Despite its decision on beer, the 1949 Committee recommended ‘not [to] exclude any group of commodities or services because that particular class of living expenditure was regarded as non essential or socially undesirable.’  By the 1955 revision, Fraser had passed on and beer was included.
In 1949 the regulations went a step further: when making a standard wage pronouncement or a general wage order , the Court was required to take into account – among other things – ‘[any] rise or fall in retail prices as indicated any index published by the Government Statistician.  The term ‘cost-of-living’ has been replaced by the more specific ‘retail prices’ – reflecting the title of the official index (even though it covered non-retail consumer prices, such as rents). It is also significant that here, and indeed in other legislative examples, a specific index – such as the ‘Retail Price Index’ or the ‘Consumer Price Index’ – is not named in the legislation. So for the first time since 1920, the Government Statistician was a witness in the Court interpreting and defending the quality and integrity of the Consumer Price Index. 
Steadily the public came to trust the CPI although there have always been outbreaks of distrust and misunderstanding. I recall in the double-digit inflation period of the 1970s people seriously (I think) arguing along the lines that since bread had gone up 10 percent, butter had gone up 10 percent and jam had gone up 10 percent, inflation must have gone up at least 30 percent.
A more valid criticism was that the average regimen does not always reflect the spending patterns of a particular group. A particularly vociferous lobby were the retired; eventually an index specific to their needs was constructed, but – to the lobby’s shock – it rose more slowly than the CPI; not surprisingly for while prices tend to move together, housing prices which rise faster than the rest. Because they had their own housing without a mortgage or were living in subsidised rental housing, the elderly spending underweighted housing expenditure relative to the average household and their price index rose more slowly. Eventually the index was dropped. 
Even when Court hearings were abandoned, the CPI continued to be used in award wage determination. Typically the union negotiators would make a demand which began with the CPI increase and then added other factors; meanwhile the employers would reluctantly concede the consumer price increase while their other factors would suggest a less generous wage increase. Rarely was the veracity of the CPI challenged, even though in the double digit 1970s and 1980s it was by far the largest element in any negotiation. (There was an associated dispute as to the extent to which wage increases caused price increases – the ‘wage-price spiral’.)
A consequence of the trust was that the CPI has been frequently used in situations where a different deflator might be have been more appropriate (although never, to my knowledge, by Statistics New Zealand). For instance in the double-digit inflation era, construction contract payments were indexed to inflation. Perhaps it was judged that a specific index would have been too expensive to construct, with the possibility that there could be subsequent litigation over it.
Another misuse has been deflating nominal aggregates with the CPI, the components of which are not the same as those in the CPI regimen. For example, deflating nominal GDP by the CPI is either brave or reflects a misunderstanding of both measures. Yet deflations like this happen too often. (Conceptually it is equivalent to analysing a single commodity economy which requires only one price; it is rare that economies can be treated so simply.)
Another problematic use is the conversion of a past nominal price to an equivalent price today using an inflation calculator. Statisticians are well aware of the difficulties (not least the introduction of new products which were unimaginable then – a motor car in 1870?). But the changed construction of the measures adds to the difficulties. (Personally I am inclined to compare the price with the wage of an unskilled labourer, well recorded since 1894, or a value with – a much shakier – aggregate nominal GDP.)
The CPI’s role in wage fixing has diminished in a low inflation regime where other factors become more important (although outcomes are still checked against it). However such is the public’s acceptance of the CPI’s veracity it became used as a target for monetary management.
Again the statute – the Reserve Bank Act – does not specify a particular index, but requires the maintenance of ‘price stability’. However, the Policy Targets Agreement between the Governor of the Reserve Bank and the Government has set limits within which the CPI is to track.
It could be argued that the choice of consumer prices in the target agreement follows neither from economic logic nor commonsense. After all, private consumption is only about three-fifths of (non-financial) market economic activity, and about 40 percent of that consumption are imports with prices over which New Zealanders have little influence (other than via the exchange rate).
I have never seen an account as why the CPI was chosen. Maybe there were two major reasons. The first was that the index was still considered important in wage setting, and there remained a belief that the wage path had the potential to disrupt price stability. (But that is true for other mechanisms such as those involving asset prices, the exchange rate and business markups.) One factor may have been that it is not revised. like say the national accounts, because all the required data is available at the time of the first release – an advantage of using a Paasche weighted index.
The second was that the CPI was so built into the public’s psyche about inflation and the public had so much trust in the index that it was natural to use it to anchor the price path of the economy. I have argued that a better anchor would be a price index which was more comprehensive of production in the economy and was less influenced by the exchange rate. But whatever the weaknesses of the CPI, it is the price measure that the public trusts, and that consideration is probably decisive – although a ‘snake’ of various price indexes might be worth considering if the objective is to improve macroeconomic performance.
In order to increase the integrity of the data base, the Government Statistician is one of a handful of public officials who is given specific statutory independence. Public trust is vital for the collection of data – most notably, but not exclusively, the census – as well when it is used. Were there less trust, providing it to the same standard of excellence would be more expensive.
Even so, government spending can influence the Government Statistician’s actions. The CPI is probably protected because of its public reputation – although of course there is a constant effort to reduce the cost of collection; the prospect of collecting from electronic price bases looms. However demands for other statistics have often been unfortunately delayed by fiscal considerations; a set of comprehensive National balance sheets will be a classic example of a data base that will turn up sometime after it is desperately needed.
What this paper has shown is that while the statutory independence is important, the trust – in the specific case of the Consumer Price Index anyway – has been accumulated over the years by a myriad of practical decisions. Not all of them have been ideal, and sometimes they have been deficient. Even more importantly, past decisions have reflected particular circumstances ranging from need to availability.
Such challenges will not go away; Statistics New Zealand’s preoccupation with maintaining the trust of the public – and the independence that is a part of that trust – will continue to be pressured in the future by the factors similar to those it has met in the past.
 My much valued personal copy was given to Hight by McIraith ‘With the Author’s Compliments’. Wolf Rosenberg subsequently acquired it and in the course of time gifted it to me. From various hand-written additions to the tables Hight may have had my copy with him during the sittings of the 1912 Royal Commission.
 J. W. McIlraith (1913) ‘Price Variations in New Zealand,’ Economic Journal, 23(91) (September 1913) pp.348-54 and (1914) ‘Contribution to Current Topics,’ Economic Journal, 24(94) (June 1914) pp.341-2.
 Reported in Wanganui Herald, (23 May, 1908) p.4. I am grateful to Sharleen Forbes, Corn Higgins, James Keating and Evan Roberts for drawing my attention to this. See also the 1908 New Zealand Official Year Book, p.541-5.
 Unless otherwise sourced, all statistics used come from New Zealand Official Year Books.
 N. S. Woods (1963) Industrial Conciliation and Arbitration in New Zealand, p.95.
 F. W. Rowley (1931) The Industrial Situation in New Zealand, p.105.
 M. N. Arnold (1983) Consumer Prices: 1870-1919.
 Woods (1963) op. cit. p.97
 Woods (1963) op. cit. p.101.
 G. A. Wood (1957) Progress in Official Statistics: 19840-1957: A Personal History, p.63-72.
 M. N. Arnold (198?) The Treatment of Housing in New Zealand Consumer Price Indices: 1919-1981.
 A view strongly conveyed by Jack Baker in his second year statistics lectures.
 IC&A Amendment Act, 1936, Section 3. The next subclause requires the rate tobe sufficient to enable the wage earner to maintain a wife and three children in a fair and reasonable standard of comfort, codifying earlier Court decisions (although they were often involving two children), and the Harvester decision of 1907.
 B. C. Ashwin interviewed by John Henderson March 1970 (Ashwin papers); the details may not be precisely correct.
 Wood (1977) op. cit. p.94.
 Wood (1977) op. cit. p.96..
 Wood (1977) op. cit. p.96.
 Amendment No 14, gazetted on 17 February 1949.
 But recall he had been before the Court as early as 1920.
 I was asked to criticise the index as biassed. Those who contacted me had no evidence for this, but believed that if it reported slower rises it must be.