This was a part of the (Long) December 1992 version of In Stormy Seas which was never published.
Keywords: Statistics;
The problem of the 1977/78 downturn arose in a different way again. In the late 1980s advocates of the policies loosely called “Rogernomics” were wont to justify their proposals by referring to the poor economic performance in the preceding decade, quoting the GDP growth statistic. We shall see in Chapter 5 their account of post-war growth was confused. Nevertheless, the conscientious had to check out their claims.
It became evident that some of the poor performance occurred by a judicious choice of beginning and end points of the period (leaving out years of high growth), but that there was some truth in the story. It was also evident that the 1977/78 year played a key role in keeping the overall average growth rate down.
In that year the official statistics recorded a 2.7 per cent contraction in the economy, a sharper fall than any other year recorded up to the time of writing. To give some idea of the size of this effect, suppose that for nine years of a 10 year there was a 1 per cent growth rate on average, and the tenth there was a 2.7 per cent contraction. In that case the decade average would be .6 per cent p.a., almost halving the growth rate.
Did the economy experience a record contraction in 1977/78? There is no doubt that some year has to be the record, but I was puzzled about the size of the 1977/78 decline. Especially as employment had risen .5 per cent for the year, somewhat inconsistently with a record decline (NZYOB 1981:679).
A check of the sectoral contribution’s to GDP suggested that manufacturing and domestic trade were key to the contraction. Other smaller sectors also contracted, but the official statistics record manufacturing net output decreased 6.2 per cent, and domestic trade decreased 5.5 per cent. Given that each contributes about 20 per cent of GDP, the two sectors explain 2.4 percentage points of the total contraction.
Frustratingly, the wholesale price index series was terminated in 1978, and a new producers price index series introduced. While there was some overlap, the change of series meant one could not check the change in the following year. Nominal series for manufacturing and domestic trade are available. Checking them suggests the problem was either in the price deflators, or an endpoint problem to be explained below in terms of inventories. The best course seemed to be to examine contemporary accounts.
The first source was the July 1978 budget speech. Here is an extract.
“By early 1975, New Zealand’s terms of trade had declined by 43 per cent as against the peak in 1973. … Since the end of 1975 we have been adjusting to this grave decline in our terms of trade. The present Government on taking office was faced with the fact that business confidence was excessive having regard to our real economic difficulties. It took nearly two years for that confidence to be blunted, and realisation of the situation did not come until the third quarter of 1977. At that time the manufacturers realised that they had been producing in excess of the ability of the economy to absorb their products. A stock overhang had resulted and the downturn in production with consequent laying-off of staff was much greater than it need have been had the real situation been understood sooner.” (1978:5) [1]
The two cyclical indicators – Haywood & Campbell reference cycle, and median manufacturing capacity utilisation (CUBO) peak in March (Kay 1983) – support this view. In neither case is the fall particularly sharp or deep. Indeed CUBO reports a decline in manufacturing capacity utilisation from 88.3 per cent in 1976/77 to 86.4 per cent in 1977/78,[2] a fall of 2.7 per cent in CUBO, which does not obviously tie in with the 6.2 per cent fall in the estimate of manufacturing production. The quarterly survey of retail sales also peaks in the March quarter. [3] There was 5.5 per cent fall between 1976/77 and 1977/78, which is consistent with the sectoral net output estimate.
Another contemporary source is the NZIER Quarterly Predictions. Looking through the forecasts produced each quarter we observe they predict stagnation, or a slight decline of volume GDP (in the 0 to -112 per cent range) for the 1977/78 year up to their November 1977 issue. Then in March they dropped the forecast/estimate to a decline of 1½ per cent, suggesting that the decline was somewhat sharper in the second half of the year. This 1½ per cent decrease was maintained for the June 1988 estimate, but in September 1988 the estimate for the volume GDP was dropped to a 3 per cent decrease, near enough to the actual outcome.[4] (QP Sept 1978:30) The reason for this second dramatic drop was explained:
“The revision was necessary because of the reassessment of company income and the much lower estimate of stock change in the economy. [5] (1978/2: 11) … Stocks increased in value by $490 million to the 12 months to March 31, 1978. … A first comment on these figures is they are considerably lower than estimated in our last issue. ($650m . ..). At that time, wholesale stocks were not yet available for the March quarter. As it turns out, wholesalers were far more successful in reducing their stocks to turnover ratio than we then estimated. Wholesale stocks (as measured in the revised series which has just commenced publication) increased by 13.3 per cent [6] … This alone accounts for a difference of over $100 million. Reasonably up-to-date estimates of manufacturers’ stocks are available for the first time in nearly a year, …” [7] (1978/2:18)
The message is that the data base is in deep trouble, with two of the three inventory series involving a major revision. Moreover, inflation was high, which means that the price effects were complex – especially on stocks levels, and even more so on the change in stock levels adjusted for inflation, which is the relevant national accounting concept.
In fact the evidence now available does not suggest that inventory declines were significant. In the QSBO manufacturers report there were substantial gains in stock levels in the June, September, and to a lesser extent December 1977 quarters, with a steadying in March 1978, while merchants report a slight decline over the year. [8] The Reserve Bank volume expenditure estimates suggest economy-wide inventories rose mildly during the 1977/78 year in volume terms, and it was not until 1978/79 that there was a fall in their level (Grindell 1981: 16). This is consistent with the QSBO reporting.
Given the account in the QP extract over the breaks in two key inventory series during 1977/78, it is possible that the NZIER and the Department of Statistics got their linking wrong, [9] and that the stocks accumulation especially for manufacturers raw materials has been underestimated. A five per cent error on manufacturers inventories would change the GDP volume of over 1/2 per cent. [10]
The RBNZ series suggests the GDP decline was driven by a decline in private fixed investment, which fell 19.5 per cent in volume terms. This is consistent with the QSBO reports of a major cutting back in investment plans throughout this period. Private consumption fell only 2.3 per cent in volume terms. [11]
What then do I think happened in the period from March 1977 to March 1978?
The economy was known to be weakening even in late 1986, with falling investment, slowly adjusting – as the 1978 budget said – to a major decline in the terms of trade, an over expansionary stance a few years earlier, and a weak world economy. But the sharp rise in consumer inflation in the June 1987 quarter cut back private consumption, and left merchants, importers, and manufacturers with rising stock levels. (They were not to regain control of their inventories until 1979.) This led to cuts in production, and a contraction in GDP. My guess is that the contraction was more like 1 to 1½ per cent, than the record 2.7 per cent. [12] That would make it the first contraction since the (recorded) fall of 1 per cent of 1967/68. Thus I have a flatter cyclical downturn in the 1976 to 1978 period than the data records, perhaps also with less recovery in 1979/80. However the contraction would still be deeper and longer one that any time in the post war era. [13]
The outcome of all this? Hopefully the reader has obtained some idea of the complexity of examining a single statistic, and seen that to do so involves a synthesis of statistical analysis, economic theory, and historical investigation.
There is a second, simpler, lesson here. I have suggested that the 2.7 per cent volume decline statistics could be inaccurate by over 1 per cent. Let me share a secret. Having worked with GDP volume statistics for many years, I habitually think they may be accurate to only plus or minus one per cent of the published figure. We must not get too carried away with the exact figure, but make sure any argument is tolerant to such a margin of error. Colleagues in the Department of Statistics would nod in agreement.
Notes
[1] In fact the decline probably began not later than June 1977, although the account here suggests that the contraction began in the September quarter. For instance according to the Quarterly Survey of Business Opinion, a net 9 per cent of manufacturers and 31 per cent of merchants report a decline in new orders in the June quarter, rising to 40 per cent and 40 per cent respectively for September.
The QSBO data used is tabulated below. All the statistics are in “net respondents reporting increase” form, except CUBO which is the median percent of utilised capacity.
Quarters (M76, 176, S76, D76, M77, 177, S77, D77, M78, 178, S78, D78, M79)
Economy Wide
Expected new building investment approvals (-12, -13, -10, -19, -23, -23, -31, -36, 35, -24, -18, -22, -14) over next twelve months.
Expected new plant and equipment approvals (-14, -3, 1, -2, -5, -12, -28, -31, -31, 18, -7, -2, 0) over next twelve months.
Numbers employed (-6, -8,- 13,- 6,- 7, 8, – 25, -35,-32,-28, -18, -4, 0) overpast three months.
Manufacturers & Builders
Stocks – Raw Materials (-20, -11, -15, 4, 4, 23, 22, 7, -4, -14, -14, 7, 1) experience during the last three months.
Stocks – Finished Good (0, 0, 4, -1, -3,21,39, 22, 9, 8, 3, -9, -3) experience during the last three months.
CUBO (88.9,87.2, 88.0, 88.8, 89.4,87.8,86.5,84.4, 84.6, 84.5, 86.7, 87.2, 87.8) at time of survey.
Merchants
Volume of Stocks (-21, -22, -13, 2, -3, -6, -6, 6, -5, -16, -25, -33, -17, -32) experienced during last three months.
Sources for this data are the original Quarterly Surveys of Business Opinion, but summaries of some of the data are available in Bowie & Easton (1987), with Kay (1983) for CUBO.
[2] As the variables apply to end of the quarters rather than quarter midpoint, the averaging involves a one eighth of the March quarters at either end, and a fourth of the quarters between.
[3] After seasonal adjustment and in constant prices.
[4] The first official provisional estimate was not available until June 1979, fifteen months after the end of the period (and 21 months after the period midpoint). Today a volume GDP figure is available about four months after the quarter ends.
[5] A reduction in the value of stocks would typically be offset in the company balance sheet by a reduction in company profits.
[6] ie. lower than the rate of inflation.
[7] The text goes on “following publication of results from the new quarterly Manufacturers’ Survey up to March 1978. Making some approximate adjustments to link to the old survey, it appears that manufacturers’ stocks (excluding primary processing) increased by about 17 percent from March 1977 to March 1978. The increase looks to have been concentrated in finished goods. (Work in progress is aggregated with finished goods in the new survey, instead of materials as previously, so the old and new series are not comparable).” (1978/2: 18)
[8] A complication is that manufacturers are not specifically asked to distinguish value from volume in their stocks report. Merchants are.
[9] The item about stocks in the 1978 budget speech suggests that the Treasury was having similar difficulties to the NZIER.
[10] .7 percent more precisely.
[11] It is not immediately obvious though how these changes appeared in sector output. The official statistics report a small rise in the construction sector of .6 per cent in volume terms. Private investment is supplied by manufactures and domestic trade, but not enough to depress their sectors as much as reported. On the other hand a major source for investment is imports and they rose (a rise supportive of the likelihood of overstocking). The small fall in private consumption probably fed back onto lower manufactures (particularly durable goods suppliers). Even so I am left with the impression that the estimated sectoral declines in manufacturing and domestic trade are too large, and the error may be attributed to the problems of linkage of old and new inventory figures.
[12] Any sharp contraction was too late in the year to drop GDP as much as the estimate suggests.
[13] And what about the .5 percent rise in employment? Well, weekly hours per worker fell from 36.9 in 1976/77 to 36.6 in 1977/78, or a 1 percent fall, so that total labour hours fell.