1. How far apart should the contracts be spaced? 3 months, 6 months, 12 months?
2. Should the contract payout be proportional the level of NGDP or the rate of change of NGDP in some period?
3. How far out should the contracts start? What are the most important expectations? Expectations about NGDP 6 months out? 12? 24?
4. What BEA data should the contracts be based on? Advance, Primary, Secondary or Final NGDP estimates?

Our latest question is as follows:

It's important to explain it a little. There are several ways in which we can look at NGDP. Using the final estimate of 2012 Q4 we see the following (using Tables A1 and C1 - for our guide on UK National Accounts see here):

Level path (YBEU): 110.8

Percentage change, quarter on previous quarter (IHYN): -0.3

Percentage change, quarter on corresponding quarter of previous year (IHYO): 1.5

Absolute amount (YBHA): £388,350m

The reason we are interested in NGDP is because we are interested in NGDP level targeting. As Lars Christensen has pointed out in relation to our previous prediction market, it's the level of GDP that matters. But as Britmouse added, the level gets routinely rebased. That's not the only problem. From a policy perspective, we may be interested in how actual NGDP relates to, say, a 5% NGDP level target. But we'd then need to choose a starting point, and compile shadow statistics that project a 5% path on top. It would be possible to project 5% NGDP growth from the height of the boom onwards, and then operate a prediction market in terms of the absolute amount of NGDP with those projections in mind. However the reason we've decided against this is that (i) it involves an arbitrary choice of target growth and starting points; (ii) it would be hard to relate those values to the figures released by the ONS. In other words we want to focus squarley on the official ONS figures.

There is a widespread discussion about GDP growth and we want to tap into that. If you ask someone what's happening to GDP they're not going to be thinking in terms of levels (relative to 2009, or some other base year) or indeed the absolute amount. It's growth rates that are in the news. So the real decision is which growth rate - should we compare it to the previous quarter, or to the same quarter of the previous year? Let's look at the data:

The headline growth rate of GDP is in terms of quarter on quarter change, which in our case is the green line. The orange line shows the percentage change based on the same quarter of the previous year. To be clear, the Q4 2012 figure in the chart differs because the green line is relative to Q3 2012, whilst the orange line is relative to Q4 2011. Given how I've argued that we want to tie into the public awareness of growth rates it'd make sense to focus on quarter on quarter change (green). But it's not very interesting. The change relative to corresponding quarter (orange) tells a far richer story, with NGDP growing by around 5% from 1998 - 2007, collapsing to less than -4% in 2009, rebounding in early 2010, and then embarking on a downward trend. Critically this seems a closer fit to thinking in terms of an overall 5% GDP level target. To be sure we're not looking at the level path, so we can't tell from the chart above what NGDP "should" be now in order to get back to the 5% path. But if you want to use 5% as your benchmark you can easily see whether we're "catching up" or "falling behind". And indeed if you advocate a different NGDP level target (for example 2%) then you can still get meaningful insights from a prediction market focused on growth rates. 

So we're very much asking a question that is imperfect. But I want to stress that the purpose of the prediction market is to generate interest in NGDP expectations. It is *not* to serve as a policy tool. If you wanted to adopt an NGDP level target you'd need a far richer suite of institutional designs. You'd need the government to release NGDP indexed bonds. You'd need to agree on a level target. Etc. This prediction markets is intended as a very small step to contribute to the wider public debate. So focusing on an annual growth rate is seen as the most modest option. We can't switch debate from real GDP QoQ growth rates to nominal GDP level path relative to an arbitrary growth rate in one step. 

The reason we're using the primary release is because this is when there's most public interest in the figure, since it provides the first look. Of course it is likely to be revised in subsequent releases, but it makes sense to only focus on one. In addition the ONS intend to start including NGDP figures in their primary release. We've decided to exploit this happy coincidence.

There are several different ways in which we could ask the question. Previously we used a multiple choice answers to deliver a real time estimate of the probability of each possible outcome. Here were the options:

As it turns out, the second estimate (released February 27th) gave a figure of 109.8. But the third estimate (released March 27th) revised this to 110. The problem isn't just ambiguity about which release would be used, it is because if the "true" value falls close to the border between different options, you lose a significant part of the predictive power. It places a higher burden on the question design (and indeed the question designer!). With that in mind this time the question works in a slightly different way. We set an opening price of £50 and this is scaled by 0.1 to imply an expected growth rate of 5.0%. Traders can either buy or sell shares relating to that probability value, and in doing so move the market. 

It remains to be seen whether there's enough interest in this market to warrant a separate one for every quarter. I suspect that we will open up a market for 2015 Q1, for 2016 Q1, etc. We'll respond to market demand. For now though jump in and start trading. If the Bank of England's Monetary Policy Committee incorporate the insights of this prediction market as a complement to their current sources of information, decision making would be improved. Indeed if they set a rule by which they would alter make interest rate decisions such that this market stays at 5%, we'd be getting somewhere...

The table below shows a stable link for the three different releases. By finding the current month, you will see wherabouts the ONS are in the cycle. The Preliminary estimates tend to get more headlines because they are the first indication of the shape of the economy. So April, July, October and January tend to be when GDP figures are in the headlines. But these are subject to revision, and it's quite possible that what gets announced as a "recession" turns out not to be one.

Also, don't forget that even the "final estimate" is subject to revisions. For example, when the Final estimate of the Q2 data is released in September we tend to compare it to the Second estimate, released in August. But the September release may also adjust the data for previous quarters, such as Q1.

When you click on the link you will have two options:

1. If you want to view the data:
   • Statistical bulletin > Download PDF
2. If you want to download the data (or see data over a longer period):
   • Reference Tables > Quarterly National Accounts Data Table

The preliminary estimate doesn't contain the same amount of information as the second and third. I believe that this will soon change, but for now:

• Preliminary estimate
  • See Table B1 - the second column from the right shows "Gross Domestic Product at market prices"
  • The headline figure that gets reported in the news is the quarterly growth rate compared to the previous quarter - series code IHYQ
• Second estimate and third estimate
  • See Table A1 - this shows annual levels, quartely levels, annual change, quarterly change (compared to previous quarter), and quarterly change (compared to previous year). The headline figure has the same series code as in the preliminary estimate - IHYQ. The third column is showing Real GDP (i.e. GDP at market prices).
  • We can also see estimates for NGDP (i.e. GDP at current prices) in the first column. 

  Note that:
  

  NGDP = Real GDP + GDP Deflator
  IHYO = IHYR + IHYU

  You can also do this for annual figures (YoY change)
  

  IHYM = IHYP + IHYS

  Or quarterly figures (QoQ change)
  

  IHYN = IHYQ + IHYT 

  • For full details of NGDP see Table C1. This shows "Gross Domestic Product at market prices" but measured in current prices. 
    • The comparable figure to the headline GDP figure is the "percentage change , latest quarter on previous quarter" - series code IHYN
    • The figure that ties in best to NGDP targets is "percentage change , latest quarter on corresponding quarter of previous year" - series code IHYO
  • Table A2 can be useful because it shows absolute levels - if you want to verify the growth rates you can calculate them yourself from this data
  • Table F shows Gross Fixed Capital Formation which is essentially investment

For more, here's a guide by the Chief Economist of the ONS.

An alternative way to find annual NGDP data is to look at "Money GDP" from HM Treasury, which goes back to 1955. 

Exercise

On April 25th 2013 The Guardian wrote:

George Osborne has welcomed news that Britain's economy expanded by a stronger-than-expected 0.3% in the first quarter of 2013, avoiding a triple-dip recession.

As the first estimate from the Office for National Statistics showed that a healthy performance from the services sector helped GDP growth to beat the 0.1% expected by City pundits, the chancellor said it was evidence that the coalition's policies were helping to "build an economy fit for the future".

Using the information in this post you should be able to go to download that very release, and see where the reported 0.3% figure comes from. 

The 8th Quarterly Meeting will be held in Copenhagen and take a detailed look at the theoretical desirability and political feasibility of NGDP level targeting being adopted in the UK. It is an attempt to bring together a broad group of people interested in Market Monetarism.

Here's the agenda:

Monday May 13th 

12:30pm: Meet for lunch

Then: Tour of Danske Bank

Then: Visit to CEPOS for discussion

Attendance is strictly by invitation only.

The decline in 2012/13 is immediately apparent, but it's important to see that by 2014/15 spending is forecast to return to it's 2007 level. The debate is really about the the severity of the contraction since the 2010 general election, the extent to which that was an inevitable consequence of a 2009 fiscal stimulus, and the one off factors (such as transfer of Royal Mail assets) that distort the figures. What strikes me though is that I don't know any economists that would advise the government to do the above.

The chart below includes 4 scenarios:

• The purple line (scenario 1) is the arch Keynesian "keep on spending until the economy recovers and to hell with the debt!" It's what I suspect many Keynesians believe, albeit are very reluctant to admit to. 
• The blue line (scenario 2) treats the 2009 stimulus as a temporary injection, and maintains investment at its pre stimulus levels
• The green line (scenario 3) acknowledges that the 2008 - 2011 spike put public debt on a new course, and therefore mirrors the rise in order to steadily return to 2004 levels
• The orange line (scenario 4) is an ultra austerian desire to shrink the state, allow depreciation to eat up the existing capital stock, and flat line with negative net investment
• The red line (actual) is taken from the March 2013 Budget.

I'd be really interested to know various preference orderings. My intuition is that scenario 3 would be preferred to the actual by both sides of the debate. Indeed one would think that the serious debate would be between scenarios 2 and 3. 

Source: 2003/04 - 2010/11 from PSF1 of the Public Sector Finances.

• General Government GFCF - Table F of the national accounts (see here for a guide to using the national accounts, and here for our comparison of General government and private GFCF).

• Public Sector Net Invesment (excluding financial interventions) - Table PSFN1 of the Public Sector Finances (see here for Sep 2012). If anyone knows how to download these figures in Excel, please let me know!

The point I'd make is that the dramatic cuts seen made by the coalition government (i.e. from 2010 onwards) are only "cuts" relative to the time of the fiscal stimulus of 2009. When Keynesian economists advocate fiscal stimulus they are arguing for temporary deficit financed spending. Therefore by definition a stimulus will be "withdrawn" at some point in the future. The argument against a stimulus is that it may be withdrawn prior to the economy recovering. If you say "don't withdraw it until the economy recovers" then to what extent is that a temporary programme? And both sides would acknowledge that all else equal, the higher the existing debt level the more dangerous a "permanent" fiscal stimulus is. Let's not also forget that part of the 2009 fiscal stimulus brought forward capital spending due to take place in 2010-2012. It's like me asking to be paid 2 weeks early and then in a fortnight time complaining about having a pay cut. 

I'd love someone to point out errors in my interpretation and use of the data (in particular I'd love to find a stable link to these indicators). I just think that although analysis like this is far more sophisticated than my own, to only start looking at what's happened since 2009/10 is deeply misleading. Keynesians - you got the stimulus you asked for. If you want another one I'm willing to debate that. But you can't simply pretend that the first one never happened!

Series code: IHYO.

It's becoming increasingly hard to argue that expectations should switch to a new lower NGDP growth rate, rather than policymakers should ensure that NGDP reflects estasblished expectations. For more see Britmouse.

P.S. Here's the forecast YOY growth of NGDP from the March 2012 OBR report:

We also draw attention to measures of private investment, which we've updated:

Addendum: In the comments section Nicolas makes an important point about measuring regime uncertainty. It reminded me that I had intended to discuss this post by Lars Christensen in the report. Lars says,

My favourite source for a numerical measure of these uncertainties is the conservative Heritage Foundation’s Economic Freedom Index. We can use the sub-index for “Rule of Law” in the Economic Freedom Index as a proxy for “regime uncertainty”.

I think this makes a lot of sense, but I find it interesting to note that (i) this measure is only calculated on an annual basis; (ii) only exists from 1995; and (iii) "property rights" is on a 20 point scale so it doesn't change much. So I'm not sure how much help it is at assigning a causal role to regime uncertainty during recessions. For example, here's the "regime uncertainty" for the US since 2004:

Lars has some other great posts on regime uncertainty here and here (written by Alex Salter).

General government expenditure can be found using the following steps: 

1. Go to the ONS release for "Maastricht Supplementary Data Tables" (the most recent one is January 2013 and can be found here. As far as I can tell there's no stable URL for them so I would suggest entering "Maastricht" into the search box of the index page for all publications).
2. Click on "Data in this release" in the top right corner.
3. Download the excel file for "ESA Table 25 Quarterly Non-Financial Accounts of General Government Expenditure".
4. Select Tab 2501 and look for the column "Total government expenditure" (possibly column T).

Here is a chart showing expenditure from January 2006 - December 2012:

Note that this measure of general government expenditure includes transfer payments, and it also includes the income from sales of goods and services. The latter are subtracted in order to generate Total Managed Expenditure (TME), a series that is not published by the ONS. 

See here for more details.

You can download the report here.