Emily Oster changed the way I think about economics

Perhaps in stark contrast to my last blog post, in this post I want to celebrate the good economics can do.

I have Eric Crampton to thank for introducing me to the wonderful work of Emily Oster. Oster's most famous book, Expecting Better: Why the Conventional Pregnancy Wisdom Is Wrong -- and What You Really Need to Know was a surprisingly engrossing read considering I have never been pregnant, have not been trying to get pregnant, and probably won't be getting pregnant anytime soon unless I end up birthing the next baby Jesus or something.

What I got out of Oster's book was a new appreciation of the economics. I learnt about the value economists can add in simply reading and evaluating the damn studies, and making sense of the data and evidence. You wouldn't think this is such a big deal. But as it turns out, in an area like pregnancy advice, it's clear that a lot of the time this just wasn't being done. And there are very real consequences. 'Conventional wisdom' based on bad studies or ignorance of better studies is everywhere when you start looking. 

Another value-add of Oster's work is that it is all about empowering individuals to think about risk and make decisions for themselves based on their risk preferences. This isn't just applicable to decisions related to pregnancy, although pregnancy turns out to be a really good example as the stakes are a lot higher and people may be a lot more risk averse than they would be otherwise. It's enabling individuals to make decisions related to their health and wellbeing, based on the best possible evidence, and knowledge of the stuff they still don't know (known unknowns). A little bit of data/evidence literacy can be a great thing!

Finally, because of my academic and career background, I tend to see economics and policy as inter-related. Obviously, this is a narrow view. There's a lot of economics that doesn't even touch on policy (and a whole lot of policy that willfully ignores economics!). But a lot of the most interesting and useful economics has to do with personal decision making: from finding a partner, to how to raise your children, to optimising your health and wellbeing, to deciding where to eat and where to live. In the absence of government telling you what to do at every stage of your life (I'm being optimistic about the future here), it helps to have analytical tools available. And economics is a very useful part of that toolkit.

Anyway, why am I talking about Emily Oster right now? Because Vox have published an excellent article on her and her career. Oster has been extremely prominent in the US media lately for her work on Covid-19 and reopening schools. As with almost any Covid-related topic, and Covid in the US in particular, the views have been considered controversial. Unsurprisingly, I recommend reading the whole thing. But here are some of the quotes I found most interesting:

“We’re facing basically this existential threat to schooling for all these kids, and public health,” she [Oster] remembers thinking, “and the best data that the CDC can marshal on this has been put together by a professor in her basement.” 

The story of how Oster has emerged as a singular authority on schools, despite her lack of a background in education policy or pandemic response, starts with an information vacuum.

This is why Oster reaches hero-status in my mind. It's easy to see an information gap and complain. Or wait for the government to get its act together. But whether you agree with her commentary and conclusions or not, the fact that individuals like her can see a massive information gap and work to fill it without access to the best resources available is worth celebrating. 

Oster says she doesn’t mind being criticized on the merits of her work, but she bristles at the notion that it’s not her place to talk about Covid-19 and schools.

 

“I find this credentialism frustrating because I think that it is an argument that people use to not bring multiple voices to a debate,” she says. “It’s absolutely true that there are a set of tools that come with epidemiology that I don’t have,” she adds. But she says she believes she has another set of tools — “thinking about risks and analyzing data and looking at evidence” — that are useful in their own right.

I think this sucks, and these attitudes prevail in NZ too. Sure, there were/are economists in NZ with bad takes on the pandemic. But that's not a reason to dismiss the analytical framework they can bring to the table. In an ideal world, epidemiology and economics would work together. But there seem to be people hell-bent on making this a war of the professions.

Oster’s trajectory over the last year is emblematic of the way a failure of federal leadership often left Americans rudderless, with seemingly nowhere to turn for guidance on life-or-death decisions. And now, as schools and the economy reopen and many are reevaluating their relationships to authority, data, and advice, Oster may be most relevant not for the times she’s tried to tell policymakers what to do, but for the times when she’s empowered people to make choices for themselves.

In the face of incompetent governments, people need the information that will help them make the best choices for themselves and their families. I'm grateful we didn't face the same problem here in NZ, but it's a lesson worth filing nonetheless: sometimes you cannot wait for the government to tell you what the right thing to do is.

Indeed, Oster is a cautious speaker — choosing her words meticulously, rarely showing emotion, unfailingly gracious to her critics.

This is who I want to be when I grow up. Too bad nearly all of these qualities are considered flaws generally, and more so when you're a woman.

Though she continued her academic work, she was increasingly seen publicly as a parenting expert (“She’s on the cover of Parenting Today; she’s not on the cover of The Economist,” Johnson notes).

Ah yes, because real economists are the crystal ball gazers and the ones who make complicated and fancy models that fail the real-world test. I'm not one to cry sexism for the hell of it. I'll let you make up your mind on that. 

Uh oh... are people only now learning economic forecasts are often wrong?

Is there a German word for the pleasure you experience when someone calls out economists who are consistently wrong? Because that's what I'm feeling. RNZ's piece on the problem with economists' forecasts reveals an issue that is not all that new, but people must have short memories. It's an issue that is dear to my heart. It's time to start calling economists out on their bad forecasts.

Full disclosure: I'm currently slowly making my way through Nate Silver's The Signal and the Noise: Why so many predictions fail -- but some don't. It's all about how to deal with uncertainty in forecasting. I'm enjoying it.

The problem with economists' forecasts is that they're often wrong. Embarrassingly wrong. 

The article gives economists a bit of a ribbing, and fair enough. But the fault doesn't wholly lie with economists -- they are after all dealing with multiple variables and layers of uncertainty attached to each. And economists presumably provide these forecasts because they're something the public and policymakers want to know. Forecasts are useful because policy responses can be slow: you really want to get your underground bunker built before the hurricane hits. Economic forecasts fill a necessary gap.

I could almost live with that, if the forecasts were delivered with appropriate humility and caveats. But often they're not -- or at least, the soundbites that make it to print do not reflect that. It's the over-confidence that bugs me. And while some of this may come down to naivety about the way media works, it's a little hard to swallow when many of these economists are seasoned media commentators.

There are also incentives at play. Economists are rewarded by the media for their grandiose predictions: they're the ones who get headlines, airtime and name recognition. By the time future events eventuate, the public memory of the exact forecast is long gone, all that remains is the memory that 'this economist said something smart on TV this one time'. The media too have an incentive to publish clear and confident forecasts: they make good soundbites. 

The RNZ article suggests the media include more caveats and disclaimers when broadcasting economists giving their forecasts. That would help. But we can go further.

Here are 3 things that I think would help incentivise better reporting of forecasts:

1) Have a public database of economists' past forecasts compared with the actual outcome. We already see a bit of this with political polling. Keeping a track record of economists' past claims has to be a good start to incentivising better commentary. Those with low confidence in their forecasts might be more reluctant to comment to media if they know they'll be held accountable. And it incentivises those who do talk to media to take more care with their claims. It's also a useful tool for the public to compare different takes and understand how forecasts stack up to reality.

2) Improve the supply-side of the equation. This is cribbed from the aforementioned Nate Silver book, who in turn references George Mason University economist Robin Hanson. Robin Hanson's recommendation is to use prediction markets, where the public can place bets on economic or policy outcomes. Hanson's insight is that there are simply very few incentives to produce good forecasts. By adding a real financial stake to forecasting, the incentives increase and shift away from just looking good to peers (e.g. being on TV and making headlines). An advantage of prediction markets is that they're dynamic: they can provide real-time information by being constantly updated when new information comes to light. Anyway, all of this sounds good and reasonable. But then I remember we don't have iPredict anymore in NZ so...there goes that idea.

3) Improve the demand-side of the equation: publish the margins of error, or confidence intervals, in forecasts. This is also cribbed from the Nate Silver book. The idea is to encourage the media and public to be better consumers of forecasts. Again, the public are already familiar with this format with political polling (whether they understand margins of error, I cannot comment). Good economists know and will admit that their forecasts carry a range of uncertainty. Being able to communicate that uncertainty is not only useful for the public, but is also fairer on the economists who should know and want to highlight the uncertainty of their forecasts.

Not all economists who make forecasts are bad, and not all forecasts are bad. What we need are better mechanisms for discerning which ones to take more seriously.

Fun things in the FPA briefings

A slew of Fair Pay Agreement (FPA)-related briefings were released last week, available at MBIE's document library. If you want to understand the complexity of the system, I recommend taking a look. There's a hell of a lot of good analysis in there.

The most important document released, the Regulatory Impact Statement (RIS), has already been covered in the media so I'm not going to rehash what has already been said. The gist, as Stuff reports, is that 'officials recommended scrapping Labour's Fair Pay Agreement plan and strengthening existing law' and the RIS highlights the significant risks and challenges with the system. Most importantly (and most damning) the RIS warns that the costs of the FPA system may outweigh the benefits. Turns out that free and frank advice isn't dead after all. You love to see it.

The briefings that were released had some fascinating bits too. The two pieces of advice that I found most interesting are: 

1) MBIE recommended a more targeted application of FPAs and pointed out the risks of not doing so; and 

2) if Business NZ hadn't agreed to be the default bargaining representative for employers then the next best option entailed significant risks to the FPA system.

MBIE recommended a more targeted application of FPAs

Remember when FPAs were sold to us as helping the poorest and most vulnerable workers in the labour market: the cleaners, security guards and supermarket workers? Unfortunately, the proposed initiation tests in the Cabinet paper for FPAs do not guarantee that these workers will be prioritised.

The threshold for what sectors/occupations get to initiate bargaining for an FPA are decided by initiation tests. Under the proposed system, the initiation test involves satisfying one of the following tests: 

1. A representation test, that would require at least 10% of the covered workforce or 1,000 employees in coverage to support initiation (whichever is lower), or 

2. A public interest test, that would require evidencing that an industry or occupation faces certain labour market issues, such that an FPA would be in the public interest. 

These thresholds seem extremely low: requiring only limited support from workers even when there is no evidence of labour market issues, and not requiring a mandate from workers if labour market issues are identified. 

But, it turns out that things didn't have to be this way. MBIE recommended having a narrower scope for FPAs and provided advice on the risks associated with the approach ultimately taken in the Cabinet paper. The previous Minister (partially) agreed to this. Here's MBIE's advice:

The briefing goes on to explain the risks of the Cabinet paper's approach of having a rather low and untargeted initiation test. First, it would make it more difficult to ensure priority for workers in sectors where there is a ‘race to the bottom’. This is pretty important if the point of the system is meant to be aimed at vulnerable and low paid workers. In fact, MBIE points out that it is the most coordinated, unionised sectors (say, in the public sector) are better placed to initiate an FPA, and could crowd out less resourced and organised sectors who are more in need. 

The lower initiation tests will also substantially increase the fiscal costs of the system. This is important given the RIS has already warned that there is a possibility that the FPA system may not produce net benefits. 

Finally, the untargeted approach makes it harder to justify the fact FPAs negatively affect human rights and international obligations. This is a biggie. The RIS covers this issues too:

Under MBIE's preferred approach, if FPAs were only able to be initiated in sectors where both tests were met, then maybe there'd be some justification for affecting human rights and international obligations. At least under MBIE's approach there'd be a stronger mandate from workers. I'm keen to see what advice comes out in the public submissions process on this point. 

What would happen if Business NZ refused to be the default employer representative?

This is a fun thing to consider: what if employers simply refused to participate in FPA bargaining? It's not really a crazy possibility, given how many employers have said they do not support the system. That's why securing BusinessNZ's agreement to be the default bargaining representative for employers if no one else steps up is so important (and yes, at the time of writing they have agreed). But I have wondered what would have happened if BusinessNZ had refused to participate in the FPA system. 

MBIE's briefing sheds some light on this. Here's the advice (excuse my bad snip and paste):

Just to reiterate although I think the advice is pretty clear: if BusinessNZ had refused to participate, the next best option has significant has significant risks, including the possibility of being legally challenged and departing from the core concept that FPAs are bargained.

The 'next best' option also comes down to how great an incentive it is to have a bargained outcome rather than a determined one. Remember, collective bargaining can be a messy and costly process. And there is still a risk of going to determination anyway if parties cannot agree. In fact, the RIS points out that:

"Although the outcome is difficult to predict we anticipate in many cases the system is likely to result in bargaining stalemates and determinations fixing terms by the Employment Relations Authority, so the added benefit of bargaining may be limited." 

So... going straight to determination may actually save both bargaining parties a lot of time and resources if they're likely to end up in a stalemate and determination further down the line. If this becomes an expectation, then there doesn't seem to be much incentive to participate in bargaining. 

Anyway, all of this is interesting because BusinessNZ have called for FPAs to be scrapped. So while the 'next best' option is currently moot, I'm filing it away in my memory palace just in case. 

To conclude, I must end with an apology: I am sorry if you got to the end of this and discovered there was nothing fun about the FPA briefings at all. For future reference, I'm the kind of person who gets a thrill out of organising my sock drawer. But even if you find FPAs hopelessly boring and complex (they are definitely the latter), the implications for freedom and the way we negotiate employment agreements in New Zealand are massive. And all this, without requiring any evidence of a real labour market problem. Boy am I looking forward to reading the submissions on this thing.

What the LTFS does and doesn't tell us about NZ Super

I'm going to take a position that's guaranteed to sideline almost everyone: we need to tweak NZ Super (NZS), but not because the long-term fiscal position statement (LTFS) tells us so.

That was the conclusion of a report I wrote a lifetime ago: Embracing a Super model: The superannuation sky is not falling. I can't bring myself to read and repeat the whole report again, but it's worth a squiz if NZS is your thing (and my God, there are a lot of people who get jazzed about Super). The main argument in that report was that the future fiscal costs of NZS are not a cause for concern in and of itself, but there are distributional impacts to consider that support the case for change.

First, let's look at what the LTFS does tell us (based on projection based on historical trends): 

• Net debt rises from 34% of GDP to 177.3% in 2061.
• Debt financing costs rise from 0.6% of GDP to 7.6% in 2061
• And Crown net worth decreases from 11.7% of GDP to -117.6% in 2061

So, fiscally we're not likely to be in a happy position if no big policy changes are made. Regardless of your views on what level of debt is acceptable, carrying high levels of debt can harm economic growth, and can leave the country vulnerable if unexpected disasters (say, an earthquake or pandemic) hit and require even more borrowing and expenditure.

The LTFS also projects future fiscal costs based on demographic change:

• The cost of healthcare rises from 6.9% of GDP to 10.5% in 2061.
• The cost of NZS rises from 5% of GDP to 7.6% in 2061.

Are these numbers big? Are they bad? These are judgements the LTFS can't make. But I'll be honest: the rise in the cost of NZS doesn't spook me.

Here's a graph from the Embracing a Super Model report which is obviously a little old now, but shows how the costs of our pension compare to the OECD average in 2050.

I find it interesting that headlines following the LTFS focus so heavily on NZS but hardly worry at all about the rising cost of healthcare. The logic cannot be that our pension costs are out of step with most other countries, nor that NZS is particularly amenable to policy change: it's the political hot potato. I'm not implying that I know of good ways to reduce health spending while achieving good health outcomes, but it's surely something to consider even if the conclusion ends up being "nope, there's zero wriggle room or opportunities to do things better".

And as for those high debt levels in the LTFS....yes they're scary, but one has to wonder whether they are actually even feasible. It assumes that spending continues to exceed revenue over a large period of time, which is a shift away from the Public Finance Act. Here's what the Office of the Auditor General had to say about the 2016 LTFS:

There are two main reasons why, from 2016 onwards, the Government's long-term financial position is projected to become unsustainable over the long term:

• government spending continues to exceed government revenue because, as a share of GDP, tax revenue is held constant and healthcare and superannuation costs increase; and
• finance costs increase significantly because all resulting operating deficits are funded by debt.

In our view, it is difficult to imagine these assumptions would hold over a 40-year horizon because both move away from many of the principles of responsible fiscal management set out in the [Public Finance] Act. The duration of these assumptions reduces the reasonableness of the outlook and potentially the confidence that users have in the 2016 Statement's main messages – particularly when we look at how government finances have moved in the past.

Yet, in spite of all of this, I still think there are tweaks to NZS that can be made. First, because where inefficient spending can be identified, changes should be made. While the universal eligibility of NZS is definitely a strength, there are ways of better-targeting it while preserving universalism. And second, because in spite of the fact that I think there's a lot to like about the NZS model, there are still tweaks that could be made to ensure the costs are better distributed: as it stands, public spending on NZS could increasingly become a tool for regressive redistribution, transferring funds to the relatively well-off. The opportunity cost must be considered where other groups face greater hardship and/or need.

We shouldn't change NZS because of what might happen in the future. We should tweak it because the model already has a regressive element and it's only likely to get worse when there is a higher ratio of superannuitants to working age people.

These are the recommendations in the Embracing a Super Model report with some further additions in italic brackets because I'm not confident the report was clear enough about this:

• Recommendation 1: Link the pension age to health expectancy (while acknowledging that there are huge variations in health expectancy both between population groups and within population groups. This will require better support from the welfare system for those people who have not reached pension age but cannot work. I think there's more work to be done about compensating the losers of this approach, but that's not a reason to avoid action IF there are net benefits).
• Recommendation 2: Index NZS to CPI only rather than both CPI and wages (again, while ensuring that those on lower incomes are not worse off under this new arrangement, with support via the welfare system).
• Recommendation 3: Contributions to NZ Super Fund should not be at the expense of paying down debt (yep, I don't have much to add to this except that it is infuriating that this isn't happening, not even in response to Covid). 
• Recommendation 4: Productivity growth will make NZS – and everything else – more affordable (this is a cute recommendation. I mean, it's right and important, but no one should be deceived into thinking this is a straightforward task).

Rightly or wrongly, the LTFS has sparked the semi-regular debate on what to do about NZS. Refusing to touch the system in order to provide certainty to those approaching the pension age isn't really an excuse: starting the conversation early means you can ensure a smooth transition rather than a sharp shock if the need arises in the future.

And, if governments (plural, because neither major party has made much progress when in government) really really do not want to even look at NZS, then we at least need to have a conversation on what does need to give. It is not likely that NZ will reach the very scary debt levels in the LTFS projections because something will need to give. It would be great to know there is a plan for what that something might be, and that this is something governments are proactively thinking about, rather than reacting to when the time comes to take action.

*One thing I would change in the Embracing a Super Model report is the denial that there is a looming fiscal crisis or that NZ is sitting on a fiscal timebomb. Overall, it kind of looks like we are if nothing changes. But the point I was trying to make is that NZS itself is not yet looking like a timebomb.