Proxy analysis

On complex issues, people often resort to proxy measures to make judgments. At think-tanks, we get it all the time. People often seem more interested in who the funders are than the time-consuming process of working out whether our arguments make sense or not.

So the authors of this week’s Australia@Work report can hardly have been surprised when Joe Hockey focused on the report’s union links. Particularly as it turns out that Hockey made his original comments afer being called by a journalist for comment on a report which he had not seen. The summary the journalist gave probably focused only on negative comments about the government, which generated the predictable response.

The actual report, however, would not immediately give any cause for confidence that it was not just pushing the union line. After all, if as its cover says it is ‘sponsored by Unions NSW’ the conclusion that its content would be favourable to Unions NSW is not exactly counter-intuitive.

In this morning’s Australian, the paper digs up a speech by Australia@Work author John Buchanan, in which he declares himself to be a socialist. Can a socialist view WorkChoices objectively?

Buchanan and his co-authors were also trying to invoke a proxy measure of the report’s worth, citing the Australian Research Council in addition to Unions NSW as a ‘sponsor’ of the research. According to The Age:

Study co-author and director of Sydney University’s Workplace Research Centre Dr Buchanan also took issue with Mr Hockey.

“(Mr Hockey) has made some very serious allegations about me and my team being union bosses in disguise masquerading as academics,” he said.

“I find the statements from the minister quite unhelpful and it is a complete betrayal of the scrutiny we receive as academics from our peers through the ARC process.”

Unions NSW secretary John Robertson said the research was jointly funded by the ARC under the stewardship of Education Minister Julie Bishop. (emphasis added)

But as the ARC was forced to point out, its funding does not equate to endorsement. The ARC approves research proposals, not outcomes. As Education Minister Julie Bishop has agreed to accept the ARC’s recommendations, she cannot even be seen to endorse proposals, let alone outcomes.

Both sides resorted to proxy measures to try to convince us that Australia@Work was, depending on proxy, good or bad research. Of course neither is a substitute for reading a 100 page plus report. Commenter Sinclair Davidson wasn’t impressed after he read it.

I think there is a lot of useful material in Australia@Work. This is mostly because it is based on a very large survey that will help our understanding of work and how people perceive it. But unlike the government, I am free to say that there is nothing intrinsically concerning if AWAs cause some wages to go down. For example, as I have argued before, penalty rates are often anachronisms and if AWAs help get rid of them so much the better.

That said, I would cast far more doubt on the analysis in Australia@Work than the underlying statistics. I have myself had a go at Buchanan’s tendentious analysis of labour market trends, and there is more of it here.

Before we are even out of the executive summary, we are told that there is an ‘alarming level of acceptance of long hours’. Only 40% of men working more than 50 hours a week want to work fewer hours it seems. But if people like their jobs, why is it ‘alarming’ that they want to spend more than 50 hours a week at them?

That people had previously been unemployed get lower wages than those who had changed jobs is put down to their ‘poorer bargaining position’, which is of course only one possibility – with less experience and lower-level qualifications being the more likely major explanations.

But it possible to trust the numbers of people with a barrow to push. As the top right-hand corner of my blog reveals, I have an ideological perspective, as Buchanan does – albeit a very different one. That doesn’t mean that either of us make up research findings.

21 thoughts on “Proxy analysis

  1. What annoys me here is (a) the double standards of some critics of Buchanan and (b) their willingness to play the man rather than the quality of the work.

    On double standards, you cannot beat Albrechtsen and the editorial writer in today’s Australian. Albrechtsen points out that “Buchanan’s study will stand or fall on its own merits” and adds that “academics Sinclair Davidson and Alex Robson” kicked off that debate”.

    Having warned us that because of Buchanan’s political leanings, one can “raise questions” about the independence and neutrality of his work, Albrechtsen omits to tell us anything about the well-known political leanings of Davidson and Robson.

    But the wider point I want to make is that true professionals can divorce themselves from their values. Until I have done my own alnalysis of the new report, I am willing to accept that this is as true of both Davidson and Robson as it is of Buchanan and his team of economists. A priori, both must be assumed to have professional integrity.


  2. “People often seem more interested in who the funders are than the time-consuming process of working out whether our arguments make sense or not.”

    I can think of one “academic”who makes it his full time job denigrating the funding of any group that doesn’t fit into his hard left attitiude. That’s what makes an argument these days.


  3. It’s not 100% clear but Davidson and Robson seem to be criticising the authors of the report for suggesting that the federal government had ‘sponsored’ it and that the consequence is ‘the likelihood that ARC funding will become more of a political football than it already is.’

    If anyone is to blame for politicising ARC funding it’s the Howard Government ministers who rushed to condemn this report on the grounds that it was partly funded by unions. Such an attack virtually guaranteed that the involvement of the ARC would become part of the public discussion.

    If the government wants to remove the ARC from accusations of party-political bias the answer is obvious: make it independent of government and let it fund projects without the need for ministerial approval. The current situation where the minister is expected to sign off on all recommendations without question is just silly. Either the minister should approve them and accept responsibility for doing so or the process should be at arm’s length and ministerial approval not required. Most people I suspect would favour the latter alternative.


  4. I agree with SD. Its just a corrupt baby-boomers club where mates give each other money. It seems more efficient just to give the money to the universities and offer tenders where there is some area the government wants some research done.


  5. ‘It seems more efficient just to give the money to the universities …’

    I don’t necessarily disagree although it would be even more a case of mates giving each other money. And there would be the problem of deciding how much to give to each university, and how to make sure they spent it on research.

    On balance I think the ARC model has a lot going for it but it’s a separate issue really to the one Andrew was talking about.


  6. Sinclair, we both know that as political commentators we show a clear bias – me to the left (although less left than Buchanan) and you to the right. Although we occasionally criticise our side of politics (as I do in a forthcoming piece on the fiscal straightjacket), we both clearly have ‘political leanings. Let’s not pretend otherwise. But my point is that as professionals writing in journal articles or preparing refereed studies we try to keep our values at arms length – how successfully is for others to judge.

    Norton illustrates my point well. Andrew has a strong set of philosophical beliefs – which come across in his occasional political commentaries. But he manages to detach himself from them in his clinical analysis of issues and facts.

    At first blush, I believe that, despite the odd emotional language (which Andrew points to), Buchanan has tried to do the same with his survey study. He has not “made up research findings”. And his political leanings are in this context irrelevant.


  7. Andrew, nice post. As you say, the main issue here is that carefully analysing research is harder than trying to paint its authors into ideological corners.

    IMHO, the ARC process isn’t bad. The paperwork is pretty straighforward (eg. compared with the NSF), and a high weight is given to publications relative to time since PhD, which seems a sensible benchmark.


  8. Fred – I agree. People would be hard pressed to know my views from academic articles I have written. The authors’ views were not criticised by Alex and myself – for our purposes those views are irrelevant. For general consumption, however, I don’t think there is anything wrong with a newspaper doing some background checks of people and their opinions – afterall that is what they are there for.

    Whether or not the authors have ‘made up’ their results is an open question. The report does not include the survey instrument, so we cannot know what questions have been asked. Nor are summary data reported. The appendix includes this statement, “Missing age data was imputed by a regression using job tenure, occupation, marital status, presence and age of children and income.” First up, where are those regrssion results? Second, I have a co-author who calls imputing data in this way as “making up the data”. (To be fair to the authors, lot of people do this.) But when I look at Appendix table 2, they don’t say how many missing answers to questions there were. They do say there were 6 partial interviews. I am intruigued by this though. The question people are more likely to not answer is their income, not their age. So how many income questions were unanswered, and what did the authors do about that. Bear in mind, the report indicates a $100 difference and, despite have no reported statistical tests of signifiance, seems to suggest that this is an economically significant difference. The point being people revealed (let’s assume truthfully and accurately) their actual income. Is this plausible? Is my asumption plausible? (No, not really).

    Another difficulty is the industry and skill classifications. Here the authors say they collected the four digit ABS codes but to get reasonable cell sizes they had to use 1 digit codes. Okay – I’ve collapsed 4 digit codes to 1 digits myself, for the very same reason. But recall this is a telephone interview – did the respondents really know their 4 digit Australian and New Zealand Standard Classification of Occupations code and their 4 digit Australian and New Zealand Standard Industrial Classification code? If the respondents didn’t know, where did the authors get the data from? Again the lack of survey instrument is a problem.

    A further problem is the extrapolation. (Why extrapolate to numbers anyway?) Where are the estimation errors? Why aren’t they reported? Why aren’t they even mentioned? Even if they didn’t have time to discuss them, we could ahve been promised them in a future statistical appendix or companion paper. Why does N in the tables vary so greatly from table to table? Why do the the items in the tables vary so greatly from table to table? Why are terms in tables not deined in the text?

    To be sure, it is possible (maybe even likely) that answers and explanations for all my questions exist and that it is all above board. Yet, the reader shouldn’t have to ponder all of these questions. When you have a lack of basic information, such as what I have indicated, and a contentious result (that is apparently contradicted in a later table) the reader is entitled to be sceptical.

    I am pleased to see that you will reserve judgement until you’ve read the document. Will you make your views and impressions public?


  9. Andrew says:
    “I have an ideological perspective, as Buchanan does – albeit a very different one. That doesn’t mean that either of us make up research findings.”

    Sinclair and Robson have basically shown these two jokers were either wrong or being mendacious. That isn’t a surprise to anyone because since the introduction of Workchoices 99% of anything posing as analysis or a proposition coming from the left has been mendaciously dishonest.

    Unions and their mates have tried to have us believe that demand curves don’t slope downwards when it comes to the labor market.

    Union advertising (propaganda) has attempted to paint the picture that unions are there to support the rest of the working public. The reality is that unions are parasitic groups that raise the wage of the members at the expense of others less fortunate.

    This analysis was a crock. It was a set up.

    Fred tries to argue that we can separate scholarly work from our political leanings. However in most cases it is the left that is trying to pretend the laws of supply and demand don’t apply.


  10. Sinkers my experience with telephone research re classification was many years ago when it was only ASIC 4 digit levels and the organisation I worked for did it pretty easily.

    Given they have used an experienced firm and understood the questions that were going to be used it would take long.

    Time to get out of your ivory tower.

    Sinkers and Alex did nothing of the sort. He asked questions which were never going to be looked by the paper.

    Incidentally one would know this quite early by the Friedman quote.
    Sinkers might’ve have immigrated here when Collective bargaining known as EBAs here started. we economic types in the financial markets were worried about wages rises ignoring market considerations. We found out employers were much more canny in understanding were the basis behind EBAS. If lower skilled wages rose then changed work process meant costs usually fell.

    Andrew, given the costs of divorce it seems to me people working over 50 hours a week and married are being quite selfish having no regard to their husband, wife and/or children.

    JC, go to an Uni library and look up the Australian bulletin of Labour. It was started by Dick Blandy and luminaries such as Judith Sloan and Mark Wooden started there.

    you then will not mix the various IR changes that have happened and gain at least some understanding of the OZ labour market.


  11. “Andrew, given the costs of divorce it seems to me people working over 50 hours a week and married are being quite selfish having no regard to their husband, wife and/or children.”

    Homer – That’s another labour market theory that doesn’t fit with the evidence. People working long hours are less likely to experience a marital break-up. More money no doubt helps, but as HILDA shows most people don’t work very long hours over long periods of time. I any case, you could work 50 hours a week and still be home by early evening, so there is plenty of time for other things.


  12. yes Andrew but that does not take away from the argument what is the person doing at work instead of being with their husband/wife and/or children.

    you seem to be supporting a very new age lifestyle. If I like then I do it no matter what the consequences are.

    Interesting that so-called lefties are indirectly supporting social conservatives


  13. Presumably earning money to support the wife and children, by meeting the needs of other people who rely on him.


  14. Sinkers and Alex did nothing of the sort. He asked questions which were never going to be looked by the paper.

    I’m trying to understand your point here, Homer. I read about the report and its results in the paper(s), and then read the report. I didn’t ask any questions of the data, apart from what the authors and consequently the press highlighted.


  15. Homer, are you suggesting we should ban husbands playing golf or going to Bunnings on the weekend? Or perhaps spending too much time blogging!!


  16. by working over 50 hours Andrew, you are having a lend of it.

    It is called selfcentredness.

    sinkers it is pretty easy you asked quaetions of the paper that was never part of what they were gong to look at hence your rather silly Friedman quote.


  17. My goodness, Homer, you haven’t looked around much at the way many people live, have you? There are plenty of marriages that only work because the couple doesn’t have to spend too much time in each other’s company – many of them are even successful in making their participants happy. Unfortunately there are also some parents whose kids are better off for them not being home all the time.


  18. Homer:

    You have often been asked to stop sending people on wild goose chases to read out of print books or periodicals that aren’t apprpriate to the discussion. The other day you were trying to send us to the Harvard Business Review to defend you argument that only collective bargaining will support higher productivity. It was obvious you had totally mixed up managment theory with labor economics.

    I don’t know if was you, but if it wasn’t you would probably agree anyway…. someone suggested that long working hours leads to family break ups. Was that you? Own up if it was!

    If that were true those wonderful socialist beacons up there on the hill such as Swden and France etc.should have lower divvorce rates. Yet in treality they couldn’t be any higher.


  19. JC,

    you have no idea of the history of the labour market in Australia.
    You continually confuse centralised wage fixing with collective bargaining better known here as EBAs and like Sinkers do ont seem to understnd how they work.

    I am giving you the best journal to read on the subject and it is right wing as well!!

    Way back when I was studying for a master’s degree what you term managment theory was termed by one economist, sorry can’t remember the name, as micro-micro economics.

    however economists didn’t want to get their hands dirty so it vanished into the ether.

    this is a great pity as Sinkers and his mate and you could profit from learning from it.

    DD, what you are deswcribing is something but it isn’t a relationship.
    Anyone who needs to work over 50 hours to do their job can’t do their job


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s