Big agglomeration gains in a growing Auckland? Or not.

A few weeks ago, when the annual regional GDP data were released, I used them as the basis for a post casting some doubt on whether we were seeing the widely-touted economic gains from the large and rapidly growing population in Auckland.

The data aren’t ideal by any means.  Among the other issues, they are only nominal, they only go back to 2000, and there are no regional hours worked data.  But they are what we have.  Since 2000, Auckland’s population –  already far and away the largest in New Zealand –  had grown rapidly, up 30 per cent, while the population in the rest of the country had increased by 13 per cent.

And yet, allowing for all the limitations of the data, GDP growth per capita in Auckland over that period had been among the lowest in any of the regional council areas in New Zealand.  Average GDP per capita in Auckland was still higher than in much of the rest of the country, but the gap had been narrowing.

nom gdp pc by region

On the face of it, it wasn’t a great advert for the success of immigration and other domestic policies centred on the belief that the growth of Auckland was the foundation of our future prosperity.

In the last few days, Peter Nunns, an Auckland transport economist, writing on the widely-read, and often stimulating, centre-left Transportblog, also used the regional GDP data, but in a post, “The contribution of agglomeration to economic growth in Auckland”, that drew quite the opposite conclusion.  His focus is on job density which, on the measure he uses, had also increased by 30 per cent since 2000.  Applying some estimates developed previously by Dave Mare and Daniel Graham (references in the Nunns post), he estimates that just over a tenth of all the (real) productivity growth in Auckland since 2000 is down to increased job density.

Nunns derives that share by estimating a real GDP per employee series for Auckland.  As he notes, the precise numbers are purely illustrative, since we don’t have regional GDP deflators.  And we also don’t regional hours worked data, and GDP per hour worked is typically a better measure of productivity.

Nunns doesn’t give any attention to how Auckland has done relative to the rest of the country of the period since 2000.  The chart above shows that nominal GDP per capita has grown less rapidly in Auckland than in most places.  Nunns focuses on GDP per employee.  Unfortunately, the HLFS employment data groupings are slightly different from those used for the regional GDP data, leaving us a smaller number of regions to compare with.

nominal gdp per employee growth

The hard to read label is Nelson/Tasman/Marlborough/West Coast combined.

Once again, Auckland has been among the more poorly-performing regions.  There may have been real and material gains from greater job density in Auckland –  as Nunns suggests –  but, if so, they must have had to offset really really weak other influences on economic performance in Auckland.

One of the commenters on Nunns’s article asked him how he responded to my earlier post, including the chart above.  This was his response (omitting the gratuitous slur on my motives etc)

Reddell’s confusing the signal with the noise. The signal is that agglomeration economies do enhance productivity. That’s an empirical fact, backed by lots of research from many places.

However, the “noise” is everything else that’s going on in the economy. That can make it hard to read the signal by looking at statistical aggregates like regional GDP. As I point out in the post, agglomeration economies account for perhaps 11-12% of Auckland’s recent economic growth – roughly one-tenth of a percentage point a year. While small gains compounded over long time periods add up to large numbers, they are often difficult to observe in the short term.

Finally, Reddell knows very well that productivity gains happen at the margin. Enabling agglomeration economies is one such “margin”, but there are a myriad of other “margins” that we should pursue. For example, New Zealand’s poor management practices inhibit productivity, as do its weak international connections and low investment in knowledge-based capital and R&D.

I don’t find that remotely persuasive.    This isn’t a single year’s data we are both looking at, but 15 years of data.  It would be great if we had a longer run of data but we don’t.  The more recent years data will no doubt be revised, but again for now this is what we have.  Over that period, in some years Auckland has done better than other places in New Zealand, and in others worse, but over 15 years there has been a non-trivial worsening in Auckland’s relative position (whether GDP per capita or per employee), despite that rapid growth in population and job density.    Frankly, his response seems (perhaps unfairly?) to amount to “agglomeration effects are real and important,  so it doesn’t really matter what the bottom line is, or how the top-down data look, I believe my model”.

And, of course, I agree with him that wealthier places tend to be denser  (but equally, as cities get wealthier they have tended to become less dense).  The issue isn’t whether the phenomenon of agglomeration economies is real, but whether those economic gains have existed in the specific instance of Auckland over the last 15 years.  There is simply no necessary reason why they should.  Density is typically an outcome of market processes (people finding it worthwhile to bunch together), but that doesn’t mean that simply pulling more people into an area by policy (which is effectively what our immigration policy does) , and then regulatorily constraining its physical footprint, will make the people in that area more productive.  If the longer-term economic opportunities in Auckland aren’t particularly good –  if, say, New Zealand impaired by distance can really only hope to compete strongly on natural resource based exports –  simply attracting more people into the confined space of Auckland could quite easily result in underperformance.  A couple of observations over 15 years isn’t remotely enough to prove the case one way or the other, but given Auckland’s underperformance over that period I think there is some responsibility on the champions of actively pursuing fast population growth and higher job density (central governments, MBIE officials, local councils, economists) to offer a good explanation for Auckland’s underperformance, not simply assert that “my model says the agglomeration effects matter in general, and hence they must have mattered here specifically”.  Perhaps there is such a good alternative story.  I would be very interested to look at it.

And, of course, there is no dispute that lots of good policies (or, typically, avoiding lots of bad policies) go into making a successful prosperous economy.  I wouldn’t agree with the Nunns list, but that is a debate for another day.

NB:  The rise in the terms of trade over the previous 15 years will explain part of the regional GDP differences, with more heavily export-oriented regions having benefited at the expense of the other regions, although much of the trend improvement in the terms of trade has been a result of falling world import prices, the effects of which should have been fairly evenly spread across the country.  Whether differential terms of trade effects are enough to reverse the rankings is another question.



Advisory “votes” on OCR decisions

Four weeks ago, commenting on his March OCR decision, the Governor of the Reserve Bank, Graeme Wheeler, was quoted telling the Herald

“I don’t think it’s a mystery that the Reserve Bank cut interest rates. In fact, we have 13 people who give advice to the governing committee who make these decisions and the advice from the 13 was unanimous.”

It was striking disclosure.  Not the fact that his advisers had been unanimous –  historically that isn’t uncommon – but the fact that the balance of advice, in numerical terms, was revealed at all, and only days after the OCR decision to which that advice related.  Neither Graeme Wheeler nor Alan Bollard or Don Brash prior to him had ever released that sort of specific information previously.  In fact, in the past the Bank has been almost obsessive about keeping secret almost anything to do with the OCR decision or the associated Monetary Policy Statement (there is a case being looked at by the Ombudsman now on just one recent example of that).  The Bank’s attitude tends to be that it is highly open and transparent about what it wants us to know –  eg the finished product, the Monetary Policy Statement –  and for the rest (“internal stuff”) it is really none of our business.  Fortunately that isn’t the law.

Following the Governor’s disclosure, I lodged this request with the Bank on 14 March

As the Governor has disclosed in the Herald this morning (page B3) the summary results of the OCR advice from members of the MPC for the most recent OCR decisions, I am requesting the same information (no names, just aggregate numbers favouring each rate option) for each OCR decision since mid 2013.

The statutory deadline for responding to that request is tomorrow (20 working days from the day they received the request).

I think it is safe to assume that the Bank is most reluctant to release this information.  It would have taken no more than a couple of hours to compile and check the information I asked for, and the Act requires (not suggests, requires) that agencies respond “as soon as reasonably practicable”.  Had the Governor’s remark to the Herald foreshadowed a new era of openness, I would have had the requested information weeks ago.  By law, I should have anyway.

We’ll see shortly whether they have decided to release some or all of the information, or to withhold it completely.  Regardless, I suspect there is some gnashing of teeth going on, and muttering beneath the breath that the Governor should never have told the Herald the voting numbers, leaving himself open to a request of the sort I have lodged.

But it is worth being clear what I am, and am not, asking for.  The request is simply for the numbers favouring each OCR option on each occasion in the previous 2.5 years.  Most of these advisory “votes” relate to decisions that occurred more than a year ago, and even the most recent request relates to the January 2016 OCR review, itself superseded by the March MPS, the results of which (and the advisory vote numbers) we already know.  All those participating are fairly senior people –  second and third tier managers, and external advisers.

I am not asking for:

  • the names of participants in the advisory group,
  • the “vote” of each participant to be identified by name,
  • copies of the one page analysis and advice each participant provides in support of his or her OCR recommendation.

All of that is official information, and could be subject to separate requests, but it isn’t what I asked for on this occasion.

Why might the Bank want to keep the information secret?  We can distinguish between their preferences and the law.

As I noted earlier, the Bank has long taken the view that the finished product is all that should be disclosed.  Anything else will only be “noise” to markets, confusing the messages, and so on.  And as it is a single decision maker model, only the Governor’s vote really counts anyway.  From within I repeatedly argued for more openness, and these were the standard lines used in response.  There is also likely to be some worry that disclosing even just the numbers favouring each OCR option, even with a lag (and recall that some of the dates I asked about are almost three years ago),  might undermine the willingness of some of the Governor’s advisors to offer advice different to the Governor’s own preference. A Governor who really didn’t want it known that a significant minority of his senior staff disagreed with him has plenty of leverage to discourage people from putting that disagreement on paper (including, but not limited to, simply tossing the person concerned off the advisory committee).

I don’t think these arguments have much merit.

First, it is now pretty common in countries with voting committees making monetary p0licy decisions for dissents/votes to be recorded, and to be disclosed pretty promptly (always with the names identified and sometimes with the reasons disclosed).  The Bank of England, the Federal Reserve, and the Swedish Riksbank are just three prominent examples.  It isn’t a universal practice by any means, but given the statutory presumption in favour of release of official information, it certainly isn’t obvious that great damage has been done to the effectiveness of policy (of policy formation/debate) in those countries where there is much fuller disclosure.

Second, it is not as if, in the nature of human affairs and especially those involving forecasts, there are many occasions when anyone is 100 per cent confident of the right thing to do.  An agency can try to present a monolithic front if it chooses, but everyone knows that behind any decision there is a weighing of pros and cons and the likelihood that some advisors will be more strongly convinced of the merits of a particular action than others.  Citizens –  and markets –  aren’t children.

Thirdly, all this information and more is already disclosed to the Reserve Bank Board, typically within days of it being offered. The Board receives all the background papers for each MPS, copies (without names) of the one page OCR advice provided by each member of the advisory group.  Board members, I’m told, can at times, and with experience, work out who provided which advice.    The Board gets copies of that material partly to assure themselves that the Governor really is being exposed to (a) good quality advice/analysis and (b) a range of views and perspectives.  If a Governor were really uneasy about it being known that some of his advisers at times disagreed with him (a) he is the wrong person for the job (in general we should be much more worried if no one ever openly disagreed with a Governor), but (b) he and his predecessors have long faced that incentive (since the Board is, after all, the group paid to monitor his performance, and able to recommend his dismissal or non-reappointment).

Monetary policy decisions have an important influence on the short to medium term path of the economy.  And they are decisions made under conditions of considerable uncertainty.  Parliament has given the power to make those decisions to a single individual, but part (a small part perhaps) of how we satisfy ourselves that he is exercising that power prudently is to be able to see, after the event, the balance of the internal advice he is receiving from his highly paid expert staff (and external advisers ).

Of course, nothing in law requires the Governor to seek formal or written advice from identified staff on where to set the OCR.  It is possible that the Bank could argue that if it is forced to reveal the advisory votes, even with a lag, the Governor would simply discontinue the practice of seeking such advisory “votes”.     But (a) the Governor has already revealed the votes on one particular occasion, with next to no lag, and (b) to do so would presumably not impress the Board, who would also lose one of their windows into the processes the Governor uses in making his OCR decisions.

But what about the law?  My request has to be decided not according to the Reserve Bank’s preferences, but under the provisions of the Official Information Act.

What are some of the relevant provisions?

First, the purpose of the Act

The purposes of this Act are, consistently with the principle of the Executive Government’s responsibility to Parliament,—

(a) to increase progressively the availability of official information to the people of New Zealand in order—

(i) to enable their more effective participation in the making and administration of laws and policies; and

(ii) to promote the accountability of Ministers of the Crown and officials, –

and thereby to enhance respect for the law and to promote the good government of New Zealand

And the statutory principle of availability.

The question whether any official information is to be made available, where that question arises under this Act, shall be determined, except where this Act otherwise expressly requires, in accordance with the purposes of this Act and the principle that the information shall be made available unless there is good reason for withholding it.

Are there any such good reasons?

The Act provides for both “conclusive reasons” for withholding information, and “other reasons”.

Take the conclusive reasons first.  Here they are

Good reason for withholding official information exists, for the purpose of section 5, if the making available of that information would be likely—

(a) to prejudice the security or defence of New Zealand or the international relations of the Government of New Zealand; or
(b) to prejudice the entrusting of information to the Government of New Zealand on a basis of confidence by—
  • (i) the Government of any other country or any agency of such a Government; or
  • (ii) any international organisation; or
(c) to prejudice the maintenance of the law, including the prevention, investigation, and detection of offences, and the right to a fair trial; or
(d) to endanger the safety of any person; or
e) to damage seriously the economy of New Zealand by disclosing prematurely decisions to change or continue government economic or financial policies relating to—
  • (i) exchange rates or the control of overseas exchange transactions:
  • (ii) the regulation of banking or credit:
  • (iii) taxation:
  • (iv) the stability, control, and adjustment of prices of goods and services, rents, and other costs, and rates of wages, salaries, and other incomes:
  • (v) the borrowing of money by the Government of New Zealand:
  • (vi) the entering into of overseas trade agreements.

I doubt the Bank would seek to invoke most of these.  Items (a) to (d) are clearly not relevant.

The Bank might seek to argue for the items under (e), since monetary policy decisions affect the exchange rate.  However, they would be on very weak ground to attempt to do so, because (e) refers to damage arising from prematurely disclosing decisions to change policy.  In other words, if I had asked for the Governor’s OCR decision after it was made but before it was released, he would have conclusive (and quite reasonable) grounds to refuse on these grounds – the decision would be disclosed prematurely.    But my request is not for information on a decision (only advice from non decision making advisers), and the request relates to advice on decisions which have already been released (in some case, several years ago).    One line of argument they might try is that the pattern of advisory “votes” might contain information about the future path of the OCR, but (a) they would have show that, not just assert it, and (b) since the decision is always the Governor’s, it doesn’t seem terribly persuasive, especially as the Bank already releases information about its own (the Governor’s) expected future path.

The Official Information Act also has a long list of other reasons for withholding, many of which I’ve repeated here.  In these cases, even if there is an argument made for withholding, that case has to be balanced against the wider public interest in disclosure.

(1) Where this section applies, good reason for withholding official information exists, for the purpose of section 5, unless, in the circumstances of the particular case, the withholding of that information is outweighed by other considerations which render it desirable, in the public interest, to make that information available.

(2) Subject to sections 6, 7, 10, and 18, this section applies if, and only if, the withholding of the information is necessary to—


  • (ii) would be likely otherwise to damage the public interest;

(d) avoid prejudice to the substantial economic interests of New Zealand; or

(f) maintain the constitutional conventions for the time being which protect—

  • (i) the confidentiality of communications by or with the Sovereign or her representative:
  • (ii) collective and individual ministerial responsibility:
  • (iii) the political neutrality of officials:
  • (iv) the confidentiality of advice tendered by Ministers of the Crown and officials; or

(g) maintain the effective conduct of public affairs through—

  • (i) the free and frank expression of opinions by or between or to Ministers of the Crown or members of an organisation or officers and employees of any department or organisation in the course of their duty; or
  • (ii) the protection of such Ministers, members of organisations, officers, and employees from improper pressure or harassment; or

(h) maintain legal professional privilege; or

(i) enable a Minister of the Crown or any department or organisation holding the information to carry out, without prejudice or disadvantage, commercial activities; or

(j) enable a Minister of the Crown or any department or organisation holding the information to carry on, without prejudice or disadvantage, negotiations (including commercial and industrial negotiations); or

(k) prevent the disclosure or use of official information for improper gain or improper advantage.

 Would disclosing the advisory “votes” on the OCR, with some lag, after release of the relevant OCR decision prejudice the “substantial economic interests of New Zealand”?  Hard to envisage, especially when there is already much greater disclosure in a variety of other advanced democratic countries.

The only argument I can see them trying to rely on is 2(g)(i), that to release the information I’m seeking would jeopardise  the “effective conduct of public affairs” by threatening the “free and frank expression of opinions” by staff of the Reserve Bank.   But, as the Bank’s own legal adviser used to tell us, officials (and especially senior ones) are expected to offer free and frank advice, and the mere fact that free and frank advice is offered is not in itself a reason to withhold the information requested.    More specifically, if I was asking for the advice of, say, Dean Ford, or Roger Perry, Willy Chetwin or Geoff Bascand individually it might be a different matter, but this request is simply for the number of votes on each side of an OCR decision.  Officials who would be feel constrained from offering free and frank advice on the appropriate level of OCR by the fact that the numbers favouring each option (jno names) would be published are in the wrong job.  Senior officials are paid to offer expert advice, and that advice is official information.  Aggregate information on the numbers in favour of each option should not be able to be kept secret, after the event, with the protection of the law.

We’ll see shortly if the Bank realizes that this is simply not the sort of information the Act is designed to protect, and makes a virtue of necessity, setting up a new protocol for the regular release of this information with a modest lag.  In fact I suspect they will decide to fight the issue anyway, relying on the delay (apparently shortening) in the Ombudsman’s processes to try to fend off disclosure for a while longer.  If so, it will reinforce a line I’ve used previously: we have a central bank quite happy to be open and transparent about stuff they know almost nothing about (what the OCR might be a couple of years hence) and not at all happy about being open (despite the law) about things they do know about, such as the policy processes they use in coming to decisions.

I will  write about the response when I receive it.  You might wonder about why I devoted so much space to the issue today.  The answer, at least in part, is to try to look at the issue in a relatively detached way, against both the statutory provisions and a former insider’s sense of what might, or might not, be strong arguments.  Sometimes the Bank’s responses to OIA requests simply annoy me, and my initial comments might be flavoured by a tinge of emotion/irritation.  In this post, I’ve tried to look for the arguments they might use, all the time half hoping that, albeit somewhat belatedly, they might finally choose to (follow up on the Governor’s one-off openness) and take some structural steps towards greater transparency.

And why does it matter?  The request was partly a matter of principle. I believe that this information should be a matter of public record, once the relevant OCR decision has been released.  But I also chose my dates carefully.  From the middle of 2013 (a time when I was still part of the advisory group) the Bank was preparing the ground for its ill-fated 2014 tightening cycle, which was only belatedly reversed over the period since June 2015.  I think we have a right to understand whether the Governor was acting alone, whether he has unanimous support from his advisers, and when any material divergences of view began to open up.  The last few years have been a somewhat inglorious episode in New Zealand’s monetary policy (and for the Reserve Bank), and the Board has done little to provide effective transparent accountability.  Fortunately, the Official Information Act is there to allow citizens to pose their own questions, with a presumptive right to access to official information to enable them to evaluate the performance of powerful officials and their agencies.