Measuring the New Zealand Transaction Sector, 1956-98, with an Australian Comparison.
This paper uses a modification of the Wallis and North (1986) method to generate estimates of the size of the transaction sector in New Zealand from 1956 to 1996, encompassing the pre-1984 period of unusually stringent (by OECD standards) restrictions and controls on the extent of market activity, followed by the liberalisation reforms of 1984-91. The ratio of transaction to `transformation' (production) employees increased quite slowly for the first twenty five years, then increases sharply in the 1980s, before stabilising in the 1990s at about 0.68, implying two workers in five occupied in transaction activities. The most striking increase is in the number of managers, which more than quadrupled. Because of this, the share of transaction employment is by now higher than in Australia.

Economic research (New Zealand)
Employment (New Zealand)
Hazledine, Tim
Pub Date:
Name: New Zealand Economic Papers Publisher: New Zealand Association of Economists Audience: Academic Format: Magazine/Journal Subject: Business; Business, international; Economics; Regional focus/area studies Copyright: COPYRIGHT 2001 New Zealand Association of Economists ISSN: 0077-9954
Date: June, 2001 Source Volume: 35 Source Issue: 1
Product Code: 8525200 Economics NAICS Code: 54172 Research and Development in the Social Sciences and Humanities
Geographic Scope: New Zealand Geographic Name: New Zealand Geographic Code: 8NEWZ New Zealand
Accession Number:
Full Text:
1. Introduction

Economists know that markets are not `free'; that the bringing together of supply and demand is itself an economic activity using real resources. And we know that real-world firms are not dimensionless `atoms' -- that profit maximisation (or whatever firms do) also requires resources devoted to coordination and control. At least since Coase's (1937) now famous article on the nature of the firm, theorists have worked on the analytical implications of non-trivial coordination and control, developing the fields of Transaction Cost Economics, Agency Theory, and Game Theory. But much less attention has been paid to the empirical side of these issues, in particular to establishing the quantitative importance of transaction costs as a user of the economy's scarce resources.

In a pioneering analysis, the economic historians Wallis and North (1986) constructed estimates of the `transaction' sector of the US economy from 1870 to 1970, measured at ten year intervals. They find that the share of GNP absorbed by transaction activities -- basically, the coordination and control of production and marketing -- begins in 1870 at about 24%, rises quite steadily to reach 55% by 1950, and remains at about that level for the next decade. This number includes the rather large proportion of US GNP then diverted to Defence expenditures (9%).

That is, if the Wallis/North methods are valid, transaction costs are indeed significant and have become relatively more so over time. These historians attribute the upward trend to the inherent properties of the process of development and modernisation, as the increased division of labour and specialisation on which improvements in production efficiency depend demands increases in the resources devoted to coordinating and monitoring activities. Such implies an interesting trade-off: to produce more goods and services we need finer partitions of the labour force, but such partitions are in themselves costly, in that they reduce the quantity of resources available for direct production or `transformation' activities.

This paper uses a modification of the Wallis/North method to generate estimates of transaction employment in New Zealand at five yearly intervals from 1956 to 1996, with a provisional extension to 1998. With some coarsening of the procedures due to data limitations, comparisons are also made with Australia, for the 1961-96 period. The analysis thus straddles two very different epochs in New Zealand's economic history: the 1938-84 period of unusually stringent (by OECD standards) restrictions and controls on the extent of market activity, followed by at first gradual then very rapid commercialisation and liberalisation of the economy, with the period of most radical reform being 1984-9(1)

In this context, we might expect to see the modernisation process showing up in rather accelerated form in New Zealand, in comparison with the US and, perhaps, with Australia, which also has liberalised its economy, but less drastically than in NZ. It is also interesting to establish the timing of the process -- in particular, the extent to which structural change preceded the post-1984 reform program.

This is what happened. The ratio of transaction to `transformation' (production) employment in New Zealand increased quite steadily for the first twenty five years from 1956, when it was 0.35, to 1981, at 0.50 -- an increase of 43%. Then there was a ten-year spurt, to 0.68 in 1991 (a 36% increase), followed by stabilisation in the 1990s. That is, by now about 40% of the work force is involved in transaction activities.(2) Compared with Australia, the transaction/transformation ratio is lower from 1961 to 1981, then overtakes and stays above the larger country's number.

Increases in transaction employment were observed in the `market making' sectors -- legal services, accountancy, data processing, finance, insurance and real estate. But the most striking increase occurred in the internal organisation of firms and other organisations -- the ratio of managers to non-clerical subordinate workers in New Zealand increased nearly threefold over the forty years between 1956 and 1996, and grew particularly quickly from 1981 to 1991, whilst its Australian equivalent showed almost no trend at all, and is by now well below the NZ number. Thus, it appears that operating in a liberalised market environment required increased managerial input within firms, as well as more resources devoted to defining, transferring and protecting property rights.

The paper is set out as follows. The next section follows Wallis & North in setting out the conceptual issues involved in dividing the labour force into transaction and transformation workers. Section 3 discusses measurement and data sources. Section 4 gives the results for New Zealand and Section 5 makes the comparisons with Australia. Section 6 concludes the body of the paper. An Appendix describes in detail the construction of the database.

2. Conceptual Issues in Measuring Transaction Costs

The methodology to be used is, with modifications, that of John J Wallis and Douglass C North (1986), who begin by noting that specialisation and the division of labour necessarily introduce the need for coordination. Specialisation means that each productive unit produces a surplus (to their personal requirements) of a small range of goods and services, which must somehow be exchanged for the surpluses of other producers to achieve a balanced consumption portfolio. So how do these exchanges or `transactions' occur?

Possible methods of coordination and control include main force (theft), chance (lotteries), and custom. Modern civil society, however, relies on two broad technologies for effecting economic transactions: price-mediated voluntary exchange in markets, and submission to established authority relationships within organisations such as firms.(3) That is, when the price and the transfer of good or service are linked directly, we observe we are in a `market' situation; when the exchange occurs `upstream' of the priced transaction -- as when a worker on, say, a monthly contract performs tasks within that month as assigned by a manager -- we are in the world of `administration' or `hierarchy', as found within firms.

To operationalise the measurement of resources devoted to coordinating exchanges, Wallis and North develop a distinction between `transformation' and `transaction' activities. Transformation is the act of adding inherent value. It may involve the physical transformation of material objects, such as combining wood and nails into the frame of a house, or spatial/temporal transformation, as when the wood and nails are delivered from the factory or shop to the building site, or intellectual transformation, as when an architect produces a plan for the house.

Transaction activities involve the transfer and protection of property rights. In the case of a house, they arise in particular when the house is sold, and include the expenses born by the buyer which are not passed on to the seller, and those incurred by the seller which would not have been necessary had the house not been sold. Wallis and North include in the buyer's transaction expenses the legal fees, financing costs, and costs of searching for houses and gathering market information. The seller has to pay real estate agents' expenses and/or commissions, and also bears the damage and inconvenience of having prospective buyers and curious neighbours tramp through the house on Open Days, and so on.

For the case of sales workers in general (of whom real estate agents are a particular example), it will perhaps help the reader if we run through the reasoning needed to distinguish transformation and transaction activities. When you, a potential customer, enter a shop, what are the sales people doing? If they are stocking the shelves and guiding you to the goods you seek; that is `transforming' -- moving things and people from where they are not useful to where they are.(4) If they are taking your payment and making sure you don't steal the stock: those are transaction activities, which would not be needed if you produced the goods for yourself.

When property rights are not being transferred, they must be safely `stored', an activity which involves the legal system, police and guards, insurance, and the expenditure by the house owner on locks and alarms and perhaps the emotional stress of lying awake at night being frightened by strange noises (the last of which will again not be possible to tease out of the available aggregated statistics). The guarding-the-stock activity can be described as preventing involuntary (on the part of one party) exchange. We call this a transaction activity because without it the system based on voluntary exchange transactions could not function efficiently.

The above discussion refers to market exchange situations between voluntary buyers and sellers. The other great class of economic transactions occurs within firms and other organisations, when goods or services are passed along a value adding chain with no money changing hands (though input services are eventually paid for) and to an extent involuntarily (by order of a supervisor). All managers, administrators and supervisors are judged by Wallis and North to be properly counted as transaction workers, who would not be needed if the division of labour did not introduce the jobs of coordination and monitoring. Clerks and data processors are record-keepers for market and bureaucratic transactions. Secretaries are complementary inputs to managers.

There may be some over-estimation here, since managers also do `creative' work such as visualising new combinations of value-yielding resources, but, on the other hand, many primarily transformation workers, especially more highly skilled technicians and professionals, have some involvement in management activities, be this as informal or part-time supervisors of other workers, or in interactions (such as meetings) with managers.

It should perhaps be noted here that `decision making' is not per se a transaction activity. Workers making decisions for themselves are basically pitting their own strength and knowledge against the laws of nature -- `if I cut these bits of woods in these places and nail them together in this pattern I will make a house' -- it is someone telling someone else what to do to build a house w rather than do it themselves -- who is involved in a transaction, as a supervisor or manager.

We will show results that include the unemployed in the transaction sector. The Americans neither include or mention unemployment, perhaps because they are so accustomed to it. But we are not accustomed to mass unemployment in New Zealand: it is a quite recent phenomenon, and it might not be proper to ignore a number which has risen from less than one percent of the workforce to as much as eleven percent in just a twenty year period. All existing models of unemployment -- Left, Right and Keynesian -- give it what in our context is a transaction cost interpretation, due to some consequence of specialisation and the division of labour. The original rationalisation of the Phillips Curve, by Richard Lipsey, had unemployment as a `structural' phenomenon caused by Keynesian downward wage rigidities impeding the rapid flow of labour between expanding and contracting industries. In monetarist search models, unemployment is voluntary and productive investment in better worker-job matches, and fluctuations in joblessness are because of misperceptions of prices and wages. In market power models, unemployment is necessary to discipline the wage claims of employed workers and their unions. In all these explanations, it is specialisation and exchange that generates -- really, requires -- unemployment. There is no unemployment in a single-household economy.

This conceptual section closes with discussion of three problematic groups: criminals, teachers, and soldiers. By definition, criminals are not `transformers' -- that is, they are not producing directly useful goods and services, at least in the eyes of society. Most crime (about two thirds) is property crime, and so could be assigned to the transaction sector, as one of the by-product costs of running a system based on formal property rights. But other illegal activities, such as crimes of passion, do not fit neatly into either transaction or transformation categories. We will not resolve this issue here.

Teachers illustrate a number of the conceptual issues raised above. To the extent that they are child-minding while the parents work, they are part of the transaction sector, facilitating the division of labour. To the extent that they are socialising the students by teaching them the norms and techniques of participation in economic life, and are contributing to more sophisticated signalling and credentialism exercises justified by costly asymmetric information, they are also performing a transaction sector activity. But to the extent they are adding to human capital that can be used in transformation activities they should (along with the producers of physical transformation capital) be classified in the transformation sector. This is quite clear in the case of teaching science and technology, less clear for the teaching of languages. As for social sciences, such as economics, teaching and research in these fields is surely directed at understanding how (specialised) inputs fit together -- surely, a transaction sector activity. This is not, of course, to say that teaching economics is not useful -- a point which deserves emphasis. Contributing to a better understanding of how the economic system functions will help towards improving the functioning of the economy, in terms of its ability to deliver goods and services that are directly valued. However, we will not attempt to sort teachers into transaction and transformation activities -- all will be assigned to the transformation sector.

Soldiers present an interesting final case. Wallis and North include expenditure on national defence, which they interpret as the costs of protecting property rights on a larger scale, in the higher of their two definitions of the transaction sector. We will follow this precedent here, though it may be true that neither Australia or New Zealand's current military resources are capable of protecting their sovereign property rights.

To conclude. The conceptualisation of the distinction between transaction and transformation activities is not straightforward, and raises many interesting, and often quite enjoyable points of definition and delineation. The best we can hope for is to establish a reasonable set of criteria which can be applied in a consistent fashion to the actual data, to which we turn now.

3. Measurement and Data

If transaction activities are exchanged for money, and if such exchanges are picked up in the national accounts, then, in principle, they could be measured from either side of the accounts as value of output or value of input. In practice, this is possible for coordination activities sold on the market, such as accounting services, but not directly for coordination services buried inside firms and other organisations, such as management supervision and clerical work.

Despite this, Wallis and North did work with an output measure, estimating the value added by transaction sector activities. For within-firm transaction activities they take the number of employees -- an input measure -- assume that transaction workers are paid the same as transformation workers within each industry, and take the resulting estimate of the wage bill as the transaction activity contribution to GNP (value added) from that industry. This is not entirely satisfactory. Some transaction workers are likely to be paid more (managers) and others less (secretaries) than the average wage. And the wage bill measure excludes the contribution of capital services, which of course are included in the directly measured value added of dedicated transaction industries.

The approach adopted here is the simpler and more direct one of working with an input measure, the number of members of the labour force involved in each type of activity. This approach can be defended not just for its computational accessibility, but also conceptually: transaction activities are an intermediate input to production, and it is the changes in the organisational technology of the economy as reflected in changes in transaction inputs that we wish to analyse.

We distinguish where necessary between what people do (occupations) and where they do it (industries), and adopt, with one exception, an `all or nothing' approach, assigning the complete contents of an industry or occupation cell to either transformation or transaction activities. The unit of measurement is Full Time Equivalent Employment (FIRE), with part-time workers converted using the Statistics New Zealand convention that one part-time worker is equivalent to half a full-time worker.(5)

All industries are transformation industries except the following. Within the private sector we distinguish four broad categories of dedicated transaction industries, following Wallis and North: Finance, Insurance, Real Estate (collectively known as the `FIRE' sector), and Business Services (which includes legal and accounting services, data processing, and guards). In the NZ numbers reported in the next section, some industries within these broad categories are assigned to the transformation sector -- Life Insurance and Architectural and Engineering Services, for example -- but in making comparisons with Australia it is not possible to separate out these more disaggregated industries, and we are forced to follow Wallis and North in assigning all business services to the transaction sector, for both countries.

From the public sector, national and local government contributes to the transaction sector the police, and its general administration functions (eg, not teachers but counting the Ministry of Education in Wellington). Defence employment is included also.

Transaction workers use some transformation sector outputs (eg office supplies) as intermediate inputs. We assign the workers involved in producing these outputs to the transaction sector of the economy, using the same procedure for both countries. From various years' Input-Output tables we compute the value of transformation sector output used as intermediate input per dollar of transaction sector gross output, and also the gross output per employee for transaction and transformation industries in total (using the employment data constructed as explained above). From these two ratios we can get an average figure for the fraction of a transformation sector worker needed to supply intermediate input to each transaction sector worker. These fractions or ratios are interpolated and extrapolated to get estimates for each of our census years.

To avoid double-counting, we must subtract from our figures for the numbers in transaction occupations those who are employed in transaction industries (eg, bank managers). Census tables giving occupational numbers broken down by industry groups are used to estimate ratios of managers, clerks and (one half) sales workers to total employment in the private and public sector transaction industries. Unemployed numbers are registered unemployed in New Zealand, and Labour Force Survey data in Australia.

All managers and clerks (which category includes secretaries) are counted as transaction occupations. As for sales workers, having noted above that they are likely to be involved in both transformation and transaction activities, and because they are quite a large group, accounting in 1996 for 8.6% of total NZ employment FTEs, we depart from the `all-or-nothing' rule and assign them 50:50 to each sector. The share of sales workers in total employment changed little over the forty year data period, so errors in our allocation will have the following effects. If in fact more than 50% of sales work is properly attributed to the transaction sector, then we will have underestimated the share of that sector in total employment, and overestimated its growth between 1956 and 1996. If the measurement bias is in the other direction, we will have overestimated the size of the transaction sector and underestimated its growth. Wallis and North simply assign all sales workers to the transaction sector.

Constructing consistent time series for numbers in the three transaction or partially transaction occupations (administrative/managerial; clerical; sales) is made difficult by periodic major revisions of the occupational classification systems. These affect in particular estimates of the number of managers. In Australia, data since 1986 in the `administrative, executive and managerial' category include `farmers and farm managers', who, for comparability with New Zealand and with earlier years should be counted in the transformation sector, and who are therefore removed from the post-1986 Australian figures.

In New Zealand the 1990 Standard Classification of Occupations included (correctly) in the managerial category many managerial and proprietorial workers who under the previous (1968) Classification had been assigned to the activity they supervised (eg restaurant and hotel managers classified as `service workers'). Fortunately, we have highly disaggregated spreadsheets of occupational employment in 1991 set out for both classification systems, from which the earlier data can be converted into the 1990 SCO format. Sales and clerical occupations also required some scaling and splicing to get time-consistent series.

The basic sources of data for both New Zealand and for Australia are their five-yearly Censuses of Population, most recently held in 1996, supplemented by Input-Output tables for various years. The quality of information deteriorates as we go further back in time, in particular with respect to the extent of disaggregation by occupation, by industry and by part-time/full-time status. The earliest years for which reasonable numbers could be constructed were 1956 in New Zealand, and 1961 in Australia. The latest reliable year is, of course, 1996, but we use some annual Statistics NZ data to provisionally estimate figures for 1998. Full details on the construction of the databases are given in the Appendix.

4. Results: New Zealand 1956-98

Table 1 shows the results from 1956 to 1996, with provisional figures for 1998. The last column gives the forty year growth ratio for each row. In New Zealand, total full-time equivalent (FTE) employment (row 13) increased by 77%, to more than 1.4 million. The rows above show that this total growth is built up from very different disaggregate experiences.

Rows 1, 2 and 3 are the numbers of people employed, on an FTE basis, in transaction occupations: managers, clerks and sales workers (of whom one half are imputed to the transaction sector in the total on row 4). The clerical and sales categories grew slightly more than the average for total employment, but it is the number of managers that demonstrates quite dramatic change, increasing more than four-fold.

Employment in the specialist transaction industries is shown on rows 5 through 8. Their share has increased, with the exception of the National Defence forces, which have been pared back from post-Korean War days. Finance, Insurance and Real Estate transaction industries' employment went up about 250%; Business Services increased more than twice as fast as this.

Perhaps surprisingly, the share of Central and Local Government administration workers in total employment increased between 1956 and 1996 -- the old `regulated' New Zealand did not apparently need as many regulators and administrators as does the present system. However, public sector administrative employment did fall in the 1990s, especially over the past few years (if the 1998 data are reliable on this).

To measure all the transaction-related work that goes on, we must add to dedicated transaction industries' FTEs the transaction workers in the transformation sector, from row 10, and the transformation employment used as transaction sector inputs, on row 11(6).

The total transaction-related FTE number is given in row 12 -- it increased by 180% between 1956 and 1996. If we add the number unemployed from row 14, and divide by the net number of transformation workers (row 13 minus row 12), we get the ratio of transaction workers (employees and unemployed) to transformation employment (row 15). We show the ratio, rather than (as Wallis and North) the share of transaction workers in total employment, to highlight the fact that, with an approximately given total labour force, every worker added to the transaction sector must be subtracted from transformation work.

The transaction/transformation ratio climbs from 0.36 in 1956 to 0.86 in 1996. In fact, the peak census year was 1991, but this observation may be unusual -- it marks a deep recession in which total employment actually dropped from 1986, whilst Business Services, Government Administration and, of course, unemployment grew quite strongly -- there may be a tendency for transaction employment to be acyclical or countercyclical, whereas transformation work is procyclical.

Without unemployment, the transaction/transformation ratio (row 16) increases less dramatically, to 0.68 in 1996, meaning that two FTE employed persons in five are now working in the transaction sector. For the first twenty five years -- 1956-81 -- the ratio grew by 43% (0.15/0.35), then it grew at a faster rate, by 36% between the 1981 and 1991 Censuses

We noted from Row 1 the particularly rapid growth in the number of managers and administrators in New Zealand. Row 17 on the table puts this figure into perspective, by dividing it by the total of all non-managerial, non-clerical employees. That is, we measure the `span of control' of managers -- the number of non-clerical workers per manager (assuming clerks to be a complementary input to management). The ratio increases nearly three-fold, from 0.06 to 0.17. Put another way, in the old New Zealand, each manager could handle about sixteen non-clerical workers; by 1996 this number had dropped to slightly less than six (and note that the ratio is steady across the three observations in the 1990s).

The size and rise of the managerial class is certainly striking enough to warrant more detailed investigation. Statistics New Zealand provided spreadsheets from the 1991 and 1996 Censuses giving very detailed tabulations of numbers employed in each 5-digit NZSIC industry by 5-digit NZSCO occupation. These allow us to examine the anatomy of the management structure in New Zealand.

Table 2 shows some data. The numbers are for total employees (full-time and part-time), and so are larger than the full-time equivalent numbers of Table 1. One possibility is that much of the growth in managers was in the prepared food and beverages sector, such as managers of takeaway food outlets, who may have just one or two other employees under their command. We see that this sector was indeed relatively highly managed in 1991, with about one manager for every four total employees. However, the sum of these comes to a bit less than 10% of the total number of managers in the economy, and so cannot be responsible for much of the substantial increase in the employment of this occupation. We can also see that total employment in this sector rebounded very strongly from 1991 to 1996 (from 58,500 to 82,002), such that even though the number of managers also went up, the ratio of managers to total employees actually slipped back, from 26% to 22.5%.

Another query often made concerns the decomposition of bureaucrats or managers between public and private sectors. Table 2 shows that public sector administration actually is relatively under-managed, at around 6%, compared to the overall average of nearly 12%, or, for example, the quite high ratio of managers to total employees (14-15%) in the private sector transaction industries, Finance, Insurance & Real Estate, and Business Services.

Finally, it is of mild interest to note from Table 2 that, in the aftermath of the 1991 Employment Contracts Act, the number of `Human Resource Managers' increased by about 25%, to more than 2,500

It would be interesting to have a finer temporal disaggregation of the numbers, especially for the decade of the 1980s, when growth in transaction activities accelerated. In particular, was the growth between 1981 and 1986 censuses concentrated in the years from 1984, and thus to be linked with the beginning of the `Rogernomics' economic liberalisations? Unfortunately, annual data on occupational employment are not available before 1985. But we do have, from Surveys of Employment, annual data on employment in the private sector transaction industries -- finance, insurance, real estate and business services.(7) From rows 5 and 6 of Table 1 above we calculate that total employment in these industries grew by 25% from 1981 to 1986. The annual data show that about 9 percentage points of this growth occurred in the first three years, or 3 points/year, and 16 points in the two years 1984-85 and 1985-86. Transaction employment growth continued to be strong over the next two years, before collapsing to two years of negative growth following the October 1987 share market collapse. Thus the transaction industry `take-off' can be plausibly linked to the financial and other market reforms.

5. Comparison with Australia

Growth in the transaction sector of the New Zealand economy seems quite striking, but is it really? Striking compared with what? We get some perspective on what happened in this country by comparing it with Australia, a rather larger neighbouring economy which, in particular, did not go through such a major commercialisation/liberalisation experience as New Zealand. Naturally, it is not a trivial exercise to prepare matching data sets for the two countries, and, in particular, it proved necessary to follow Wallis and North in the United States and assign all industries in the `FIRE' and Business Services sectors to the Transaction Industry category, in order to get comparable classifications for both countries. The result of this is that estimates of the size of the total transaction worker sector in NZ are higher than the more accurate (I believe) figures shown above in section 4.

Chart 1 plots transaction/transformation employment numbers for both countries from 1961 (the first year for which usable Australian data could be compiled) to 1996. This is what we see: the transaction ratio begins lower in New Zealand (0.403 compared to 0.434), but grows faster, and some time before 1986 New Zealand moves above Australia, reaching a 1996 ratio of 0.714 compared with 0.652. In sum, over the 35 years, the Australian ratio of transaction to transformation workers increased by 50%; the equivalent New Zealand ratio by 77%. If we included unemployment in the transaction sector, both growth and the differences are larger, because unemployment increased more in New Zealand. Australia grows by 68% from 1961 to 1996; New Zealand by 116%.


Thus we can see that a change in transaction technology does seem to have occurred in Australia as in New Zealand, but the change is less dramatic in the larger country, and is also smoother, with no sign of a spurt in 1980s, unlike NZ. Australia may have been more `advanced' or modern than NZ in the 1960s and 1970s, but New Zealand's recent changes have made it a noticeably more transaction-intensive economy in the 1990s.

What accounts for the trans-Tasman difference? Perhaps surprisingly, the share in total employment in 1996 of the specialist private sector transaction industries (FIRE and Business Services) is similar in the two countries, and the growth in these industries over thirty five years was only a little higher in New Zealand. That is, New Zealand's `more-market' reforms do not seem to have required, relative to Australia, more workers devoted to the specialist market-making and market-using industries.

So what does account for the difference? The big factor is the explosion in managerial numbers in New Zealand, which did not happen in Australia.

The discrepancy is really quite striking, as Chart 2 shows. The Australian ratio of number of managers to number of nonclerical employees starts near 0.1 in 1961, Ssags a bit, then moves up to about 0.11 in the 1990s -- one manager per nine workers


The New Zealand story, as we have seen, is quite different. The manager/employee ratio starts at 0.08 in 1961, catches up with Australia by 1971, edges up a bit more to 1981, and then sprints away spectacularly, to 0.174, by 1996. Putting this another way, the eventual difference is equivalent to about one worker in twenty being available in Australia for directly productive work in the transformation sector who in New Zealand would be occupied in management activities. Of course, we hope that these additional managers were able to deploy the reduced quantity of labour resources under their control so as to increase their productivity sufficiently for total output to rise.

We can make comparisons of the Australia and New Zealand transaction costs numbers with the Wallis and North (WN) figures for the United States. The shares of transaction employment in total employment in 1971 and 1996 are 0.33 and 0.39 in Australia, and 0.31 and 0.42 in New Zealand. Assigning 100% instead of 50% of sales workers to the transaction sector, following WN, increases these shares by about 4 percentage points. WN estimate (1986, Table 3.13) that by 1960-70, about 55% of US GNP is generated in the transaction sector. This includes 9% of GNP attributed to the Defence industries, which accounts for only about 1% of employment in New Zealand, and even less in Australia.

That is, in the civilian economy, the transaction sectors of Australia and New Zealand were substantially smaller (in relative terms) than was the United States by 1970, and were probably still slightly smaller in 1996(8), though of course we do not know what has happened to the US transaction sector ratio since 1970, except that the share of Defence in GNP has more than halved. Growth in the New Zealand transaction sector appears to have occurred later than in the United States, and faster, reflecting perhaps a late-comer's spurt towards modernisation -- the share of transaction employment in the total grew by more (61%) in New Zealand over the 35 years 1956-91 than did the transaction sector share of GDP over the sixty year period 1890-1950 in the US (57%).

6. Conclusion

To summarise the results: modern New Zealand has experienced a substantial and quite rapid change in its organisational technology, towards a more transaction-intensive market structure. This change was occurring from the beginning of the forty years of our data period, but appears to have accelerated in the 1980s, during the 1984-91 period of economic liberalisation and reform, and to have stabilised in the 1990s. The changes appear to be rapid in comparison with the path of change in the United States, as documented by Wallis and North, and substantial in comparison with Australia, such that by 1996 about one more worker in every twenty who would be doing directly productive `transformation' work in Australia would in New Zealand be a transaction worker -- specifically, a manager.

Now, it is worth again stressing that there is nothing inherently `wrong' with a shift in resources to the transaction sector. On the contrary, to the extent that productivity growth depends on specialisation, it is almost inevitable that resources will need to be diverted to coordination of the division of labour. However, we do not know enough about the processes generating changes in transaction technologies to be sure that the overall division of labour between transformation and transaction activities is necessarily going to be optimal, especially when an instrumental force in effecting change is government policy, as it was in New Zealand after 1984.

Some economists, including North and Wallis themselves (1994), have proposed that market forces will act to minimize the sum of transformation and transaction costs, but this is still an assumption not a result. Given that many transaction activities occur in the setting of Principal/Agent situations, which in turn feed on private information, or private market power, and given the sheer size of the transaction-related sector identified by Wallis and North and by these numbers for Australia and New Zealand, it would be good to have more formal theoretical attention paid to the positive and normative implications of transaction-intensive economic systems.

As for the numbers themselves, their true empirical significance remains to be uncovered. The speed and scale of the change to transaction employment does add up to a massive structural shift -- larger at least in quantitative terms than any of the better known structural changes, such as the movement of labour out of agriculture, and the increase in the share of computers in the capital stock. But this does not mean that it is necessarily important -- any more than is, say, the decline in the percentage of adult males who wear hats (another large structural change in post-war NZ). Other partitions of the data, for example focussing on the `information sector' (see Engelbrecht (1997)), may turn out to be more pertinent. Any such conceptualisations will eventually need to prove their worth empirically, by improving the explanatory power of our models of productivity and economic growth.

The author thanks three referees and participants in seminars at the universities of Auckland, Lincoln, Waikato, Edinburgh, Warwick, and Simon Fraser University, Carleton University, University College Dublin and the 1997 Industry Economics Conference in Melbourne, for their comments and suggestions. Statistics New Zealand is thanked for supplying unpublished spreadsheets of Census data, and the University of Auckland Research Committee, and the Auckland Business School for research funding assistance.

(1) See Bollard, Lattimore and Silverstone (1996) for an account of New Zealand's political economy since 1938.

(2) 0.68/1.68 = 0.405

(3) Both market and intra-firm exchanges take place within the jurisdiction of the higher authority of government and the legal system.

(4) Wallis and North include transportation activities in the transformation sector, correctly, in my opinion.

(5) We follow this convention in constructing Australian FTEs, even though (a) it appears to overestimate the average hours worked by part-time employees; (b) the cutoff points differ in the two countries -- part-time is less than 35 hours/week in Australia; less than 30 hours/week in New Zealand.

(6) The workers counted in row 10 are not double-counted in row 11.

(7) Obtained from Statistics New Zealand Monthly Abstract of Statistics, Table 5, `Employment in Surveyed Industries', various issues.

(8) A caveat in making comparisons is that the Australia and NZ numbers are employment based measures, whereas Wallis and North use GNP measures (though in fact, most of their estimates turn out to be employment-based, too). Dollery and Leong, using a methodology that more closely follows Wallis and North, report (1998, pp224-5) estimates for the size of the Australian transaction sector around 1970 that are closer to the US numbers than those shown here.


Bollard Alan, Ralph Lattimore and Brian Silverstone (1996), "Introduction", in Brian Silverstone, Alan Bollard and Ralph Lattimore (eds), A Study of Economic Reform: The Case of New Zealand (North-Holland, Amsterdam).

Coase, Ronald H. (1937), "The Nature of the Firm", Economica 4 (N.S.), 386-405.

Dollery, Brian and Wai Ho Leong (1998), `Measuring the transaction sector in the Australian economy, 1911-1991', Australian Economic History Review 38, 207-231 (November).

Engelbrecht Hans-Jurgen (1997), "A comparison and critical assessment of Porat and Rubin's information economy and Wallis and North's transaction sector", Information Economics and Policy, 9, 271-290 (December).

North, Douglass C. and John Joseph Wallis (1994), "Integrating Institutional Change and Technical Change in Economic History: A Transaction Cost Approach", Journal of Institutional and Theoretical Economics, 150, 609-624.

Wallis John Joseph and Douglass C North (1986), "Measuring the Transaction Sector in the American Economy 1870-90", chapter 3 in Stanley L. Engerman and Robert Z. Gallman (eds), Long-Term Factors in American Economic Growth, Volume 51, NBER Studies in Income and Wealth, University of Chicago Press.

Appendix: data definitional and measurement issues

A1: Occupations

The three occupational groups whose work is classified as transaction activities are Managers, Clerks and (one half of) sales workers. Censuses give information on numbers in each of these `Major Groups'. The main problem is that there were two changes in the NZ Standard Classification of Occupations (SCO), in 1968 and 1990. Splicing is required. This was dealt with as follows:

The NZSCO 90 Major Group 1, `Legislators, Administrators and Managers', appears to be defined appropriately. The SCO68 major group 2, `Administrative, Managerial' includes managers with generic job descriptions, such as managing director, general manager, administration/accounting and/or personnel manager, but excludes managers whose function links them explicitly with an occupational or industry group, such as `retail and shop manager', and `office manager', and `working proprietors' for sales and service sectors. There is a Concordance between the two Standard Classifications of Occupations which reveals that, even at the most detailed 4 or 5 digit level, there is not always a clean one-to-one map between them. However, Statistics New Zealand were able to provide disaggregated data for 1991 classified by each of the SCO systems. If, for that year, we add up (for the SCO68) all of Major Group 2 (Managerial and Administrative Workers); minor group 1391 (Head teachers, school principals); minor groups 3001-9 (Clerical Supervisors); 3101-9 (Government executive officials; 4001-9 (Managers, wholesale and retail trade; 4010-9 (Working proprietors, wholesale and retail trade); 5001-9 (Managers, catering and lodging services); and 5010-9 (Working proprietors, catering and lodging services), we get a total (full-time and part-time) of 167,196 workers, which is larger than the 1991 figure for SCO90 Major Group 1:162,291. Inspection of the Concordance suggests that many in the two clerical managerial categories (supervisors and government executive officials) were classified in the SCO90 as clerical workers. If we subtract 17% of the totals in these two categories, the sum of the SCO68 numbers equals the SCO90 figure for total number of administrators and managers.

Therefore, we perform this adjustment for the Census data from 1971 through 1986, which are all presented under the SCO 68. Of course, the workers added on to the management category are subtracted from the total numbers for Clerical and Sales workers.

The occupational classification in force before 1971 was more like the SCO90 than the SCO68, in its conceptual basis, in that it classified people by what they do rather than a mixture of what they do and where (in what industry) they do it. There is no concordance, and no overlapping data for splicing. Indeed, there were no inter-Censal occupation employment data at all in these years, so that even a near-splice is not possible. However, there is no obvious discontinuity in the series between 1966 and 1971.

Two problematic groups need discussion. `Farm managers' are classified as agricultural workers (and thus as being in transformation activities) in both SCOs. Although this will result in some underestimate of (human) managerial input, it is probably more realistic as an assumption than the other extreme of including all farm managers in the management category, because what such people are actually `managing' (especially on small farms) is mostly land and machinery, rather than workers.

The other group is (non-clerical) supervisors (including `foremen'), who are defined in the 1995 SCO manual (the 1995 SCO differs only in minor respects from the SCO90) as persons `who control and supervise a group of workers without doing any managerial tasks or duties', and are directed to be classified to the occupation they supervise rather than to the managerial category. From a conceptual transaction costs point of view, it would be better if supervisors were added in with higher managers -- their omission will bias downwards our estimates of the numbers involved in administrative work.

A2: Intermediate Use by Transaction Industries of Transformation output

The transaction sector uses input in addition to the primary labour (and capital) inputs it employs. WN attribute all intermediate input purchases by transaction industries to the transaction sector, but this must involve some double-counting, since some transaction industry output is purchased by other transaction industries. In this study a three-step procedure was followed.

First, Input/Output (I/O) tables are used for the available years to determine the proportion of total gross output of transformation industries that is purchased by transaction sector industries (for example, their buildings, desks, paper, computers, cars). Given the structural shift away from transformation and towards transaction industries, we expect two opposing forces to operate on this ratio. It will tend to rise because there will be more transaction output to be produced, and less transformation output in total from which the requirements of the transaction sector can be met. But it will tend to fall to the extent that the structural shift affects the technology of the transaction sector itself, such that it tends to use relatively more transaction and fewer transformation inputs.

The proportion of transformation output purchased directly by the transaction sector is 0.014 in 1971/72, 0.021 in the 1981/82 I/O table, and then jumps up to 0.035 in 1986/87. The most recent available I/O table is for 1992/93, and this has some changes in industry definitions. In particular, the former category `Public Administration' disappears. The new categories `Central Government Services' and `Local Government Services' must include activities previously included in the branches of government deemed to provide transformation services, as it is twice as large as it would be expected given the SIC based employment data used here. With this adjustment made, it appears that the ratio has kept about steady since 1986/87, perhaps surprisingly. The value for 1992/93 is 0.034.

The second step is to interpolate and extrapolate the I/O-based ratios to our Census years. The numbers used are 0.009 for 1956, 0.01 for 1961, 0.012 for 1966, 0.018 for 1976, 0.035 for 1991 and 0.034 for 1996.

The third step is to convert these ratios to FTEs, which is done simply by assuming that the labour/output ratio of this transformation output used as transaction sector input is the same as the ratio for total transformation output.

A3: Eliminating double-counting of transaction occupations in transaction industries

Those managers, clerks and (one half of) sales workers who make up the designated transaction occupations and who are employed in designated transaction sector industries must be netted out to avoid double-counting. Census tables on occupation major group full-time employment by Industry major division are used for this, with occupation numbers adjusted for changes in the SCONZ (as noted above). The results of this are shown on Table A1 (note that only one half of the sales workers will be assigned to the transaction sector).

A4: Unemployment

Numbers unemployed are as reported in the Censuses. From the 1986 Census, a distinction is made between `unemployed and actively seeking full-time work' (numbers = 125,094 in 1991) and `unemployed and actively seeking part-time work' (numbers = 38,676). These two numbers are simply added together, to maintain consistency with earlier Census years.

Census unemployment figures always exceed Registered Unemployment numbers, until about 1986, whereafter they move into approximate parity.

A5: Full-time equivalent employees (FTEs) by industry

FTEs are defined, following Statistics New Zealand, as the number of full-time workers plus one half of the number of part-time workers. Part-time workers are defined as those working fewer than 30 hours/week. Before the 1986 Census, the official definition of part-time work was fewer than 20 hours/week, and this was the breakdown used in the industry-level employment data. For 1986 and before, the number of part-time workers in each industry is estimated assuming that the proportion of workers in the total labour force working between 20 and 30 hours/week (for which Census data are given) applies for each industry, and the FTE number derived from that.

From the 1971 Census and earlier only total numbers employed are reported at the industry level, and FTEs are constructed assuming that the total labour force proportions of full and part-time employment apply to each industry. Note that part-time work in the earlier years is much less important than it became later, being around or less than 10% (or 5% of FTEs) up to 1976.

Employment numbers are measured in the explicitly named transaction industries, and the residual from subtracting total transaction employment from total employment is assigned to be transformation sector employment. The main difficulty here is the coarsening of the industrial disaggregation with older Censuses, especially before 1971, which Census used the 1968 SIC system. This meant that all insurance industries had to be lumped together as belonging to the transaction sector, even though it would be conceptually better to assign life insurance and possibly fire insurance to the transformation sector.

In terms of the (1987) Standard Industrial Classification, the transaction industries are:

-- `FIRE' (SIC groupings 81 (Finance), 82 (Insurance), 831 (Real Estate), excluding the category 83122, (owner-occupied housing))

-- Business Services (SICs 8321 (Legal Services), 8322 (Accounting etc Services), 8323 (Data Processing), one half of 8325 (Advertising Services), 8329 (Business Services N.E.C, which includes: Security Services, Debt Collecting, Typing etc Services, Contract Packing Services, Enveloping and Mailing Services, Management Consultant Services))

-- Government Administration (SIC Division 91 (Public Administration and Defence)

A6: Australian Data

The Australian data were compiled and processed basically in the same way as the NZ data. The sources used were hard copies of the five-yearly Censuses, back to 1961; Input-Output tables for 1968-9, 1978-9, 1986-7, 1992, and 1993-4; and two useful historical data manuals, The Labour Force, Australia: Historical Summary, 1966 to 1984, from the Australian Bureau of Statistics, Catalogue No. 6204.0, and Australian Economic Statistics 1949-50 to 1994-95, Reserve Bank of Australia, Occasional Paper No. 8 (1996).

It is of course important to achieve consistency, both over time and with New Zealand. Over time, the Australian definitions of occupations and industries are more stable than in New Zealand. The only structural break occurs between 1984 and 1985, which saw Farmers classified as Managers, and Sales workers lumped together with Personal Service workers. However, disaggregated data from the Census were available to be used to continue the time series past 1984 on a consistent basis.

As for consistency across the Tasman, no concordance is available between Australian and New Zealand occupational classification codes, but the language used is quite similar, and it seems reasonable that the Census authorities in the two countries mean similar things by the occupational categories `managers', `sales' and `clerical'.

For industrial classifications, we have now a combined Australia and New Zealand Industrial Classification System (ANZSIC), with concordances back to the earlier SICs of the two countries. From these we can see that the definitions of the transaction service industries (Finance, Insurance, Real Estate, Business Services) are in aggregate similar, with the following exceptions: The Australians include `pest control', `cleaning services' and `scientific research' in the Business Services category, and NZ does not, and the NZ definition includes `book publishing', which is placed in the manufacturing sector in Australia. Unfortunately, disaggregated Australian data are not available for earlier years, so we are unable to eliminate these differences in definition from the data. Judging by the NZ data, the net effect will be to inflate the Australian numbers for Business Services employment by about one percent of the total labour force, compared with New Zealand. This means that the Australian transaction/transformation worker ratios calculated for this study are higher than they would be if the industry definitions matched those of New Zealand.

Tim Hazledine, Department of Economics, The University of Auckland, Private Bag 92019, Auckland, New Zealand. email:
Table 1: Transaction and Transformation Employment in New Zealand,

                                               1956      1961      1966

1.  Managers                                  41244     57747     67464

2.  Clerks                                    97081    114041    140171

3.  Sales workers                             65835     66462     76403

4.  Total Transaction     =row1 + row2 +
    Occupations           0.5 row3           171243    205019    245836

5.  Total FIRE employment                     17298     19585     25471

6.  Total Business Services
    employment                                12137     14485     17047

7.  Total Gov't Administration
    employment                                21780     24898     27387

8.  Defence employment                        10037     9910      11041

9.  Total transaction     =row5 + row6 +
    industries employment row7 + row8         61251     68878     80947

10. Transaction workers in
    transformation industries                141152    169654    202732

11. Transform. Inputs to
    transaction industries                     6784      8248     11271

12. Total Transaction     =row9 + row10 +
    employment            row11              209187    246780    294950

13. Total FTE
    employment                               815035    893661   1020176

14. Unemployment                               6596      6898      9107

15. Ratio transaction     = (row 12 +
    workers/              row 14)/(row13 -     0.36      0.39      0.42
    Transformation        Row 12)

16. Ratio transaction     = row 12/(row
    employment/           13 - row12)          0.35      0.38      0.41

17. Ratio Managers        =row 1/(row 13 -
    /Nonclerical          row1 - row2)         0.06      0.08      0.08

                                               1971      1976      1981

1.  Managers                                  76954     99939    106239

2.  Clerks                                   155681    173586    180905

3.  Sales workers                             84065     89254     92955

4.  Total Transaction     =row1 + row2 +
    Occupations           0.5 row3           274667    318151    333621

5.  Total FIRE employment                     30569     37863     43766

6.  Total Business Services
    employment                                22111     26509     32661

7.  Total Gov't Administration
    employment                                33689     45971     61699

8.  Defence employment                        12686     13490     13833

9.  Total transaction     =row5 + row6 +
    industries employment row7 + row8         99054    123833    151959

10. Transaction workers in
    transformation industries                220048    250291    254349

11. Transform. Inputs to
    transaction industries                    14073     20351     23847

12. Total Transaction     =row9 + row10 +
    employment            row11              333175    394475    430155

13. Total FTE
    employment                              1104243   1254421   1287530

14. Unemployment                              16168     26337     60256

15. Ratio transaction     = (row 12 +
    workers/              row14)/(row13 -      0.45      0.49      0.57
    Transformation        Row12)

16. Ratio transaction     = row12/(row
    employment/           13 - row12)          0.43      0.46      0.50

17. Ratio Managers        =row1/(row 13 -
    /Nonclerical          row1 - row2)         0.09      0.10      0.11

                                               1986      1991      1996

1.  Managers                                 143659    156738    182172

2.  Clerks                                   202690    175587    192986

3.  Sales workers                             99407     96644    124801

4.  Total Transaction     =row1 + row2 +
    Occupations           0.5 row3           396052    380647    437559

5.  Total FIRE employment                     53659     57102     60966

6.  Total Business Services
    employment                                44095     62856     79575

7.  Total Gov't Administration
    employment                                61983     68298     63110

8.  Defence employment                        13986     12907     14289

9.  Total transaction     =row5 + row6 +
    industries employment row7 + row8        173723    201163    217939

10. Transaction workers in
    transformation industries                298151    259584    305900

11. Transform. Inputs to
    transaction industries                    42528     53732     61056

12. Total Transaction     =row9 + row10 +
    employment            row11              514402    514479    584896

13. Total FTE
    employment                              1388813   1275801   1439066

14. Unemployment                             109191    163770    152121

15. Ratio transaction     = (row 12 +
    workers/              row14)/(row13 -      0.71      0.89      0.86
    Transformation        Row12)

16. Ratio transaction     = row12/(row
    employment/           13 - row12)          0.59      0.68      0.68

17. Ratio Managers        =row1/(row 13 -
    /Nonclerical          row1 - row2)         0.14      0.17      0.17

                                               1998    96/56

1.  Managers                                 184104      4.42

2.  Clerks                                   185033      1.99

3.  Sales workers                            127578      1.90

4.  Total Transaction     =row1 + row2 +
    Occupations           0.5 row3           432925      2.56

5.  Total FIRE employment                     60156      3.52

6.  Total Business Services
    employment                                90501      6.56

7.  Total Gov't Administration
    employment                                56303      2.90

8.  Defence employment                        12748      1.42

9.  Total transaction     =row5 + row6 +
    industries employment row7 + row8        219707      3.56

10. Transaction workers in
    transformation industries                298092      2.17

11. Transform. Inputs to
    transaction industries                    61809      9.00

12. Total Transaction     =row9 + row10 +
    employment            row11              579608      2.80

13. Total FTE
    employment                              1455887      1.77

14. Unemployment                             159491     23.06

15. Ratio transaction     = (row 12 +
    workers/              row14)/(row13 -      0.84      2.42
    Transformation        Row12)

16. Ratio transaction     = row12/(row
    employment/           13 - row12)          0.66      1.98

17. Ratio Managers        =row1/(row 13 -
    /Nonclerical          row1 - row2)         0.17      2.81

Table 2: Occupational Breakdown By Industry, 1991 and 1996

1996                           human      managers   %
                               resource              manager
                               managers              s

Total Cafes, restaurants,      66         18423      22.5
hotels, etc
Total FIRE & Business          747        29142      15.0
Total Public Administration    372        6441       8.1
Total Other Services           441        26139      6.1
Total other industries         915        109125     12.8

Total 1996                     2541       189270     11.6

1991                           human      managers   %
                               resource              manager
                               managers              s

Total Cafes, restaurants,      45         15222      26.0
hotels, etc
Total FIRE & Business          432        23115      14.3
Total Public Administration    318        6396       7.6
Total Other Services           246        18207      5.8
Total other industries         948        99354      12.7
Total 1991                     1989       162294     11.6

1996                           clerical   %          total
                               workers    clerical

Total Cafes, restaurants,      4107        5.0        82002
hotels, etc
Total FIRE & Business          58929      30.3       194550
Total Public Administration    20724      26.2       79083
Total Other Services           44055      10.4       425031
Total other industries         94008      11.1       850143

Total 1996                     221823     13.6       1630809

1991                           clerical   %          total
                               workers    clerical

Total Cafes, restaurants,      3009        5.1        58500
hotels, etc
Total FIRE & Business          57747      35.8       161334
Total Public Administration    23838      28.5       83697
Total Other Services           35568      11.4       312555
Total other industries         80673      10.3       784314
Total 1991                     200835     14.3       1400400

Table A1: Ratio Transaction Employees to Total Employees in
Transaction Industries

       FIRE and Business Services   Government Administration

       Managers   Clerks   Sales    Managers   Clerks   Sales

1956     0.10      0.43    0.13      0.04      0.315     0.0
1961     0.11      0.45    0.13      0.045     0.31      0.0
1966     0.12      0.45    0.13      0.05      0.305     0.0
1971     0.12      0.48    0.13      0.055     0.3       0.0
1976     0.13      0.46    0.13      0.06      0.295     0.0
1981     0.14      0.41    0.13      0.065     0.29      0.0
1986     0.18      0.38    0.11      0.07      0.285     0.0
1991     0.20      0.40    0.11      0.075     0.28      0.0
1996     0.20      0.38    0.11      0.08      0.26      0.0

Sources: NZ Census of Population and Dwellings, Labour Force, all
years, modified as noted in text
Gale Copyright:
Copyright 2001 Gale, Cengage Learning. All rights reserved.