I thought I'd try a couple of models out on the UK economy.
First, the dynamic equilibrium unemployment rate model. This turned out reasonably well, but I am confused by a few shocks where I cannot identify a source. Here is the resulting fit:
The transitions (shocks) are at 1975.3, 1981.2, 1988.8, 1991.4, 1997.3, 2005.9, 2009.0, 2011.4, 2014.3. Most of these line up with shocks to the US or EU (labeled), representing either global recessions, or in the case of 2011.4 the ECB raising interest rates. However this leaves four shocks that are unaccounted for labeled A (1988.8), B (1997.3), C (2005.9), and D (2014.3).
The shock labeled C is the most uncertain, and might correspond to a delayed reaction to the early 2000s stock market bust. It could be left out and chalked up to measurement error or random fluctuations.
The shocks A, B, and D are all positive shocks to the UK economy, reducing unemployment. Shock D corresponds to the positive shock in the US associated with the ACA (2014.4). It is not impossible that the subsequent "boom" in the US economy boosted the UK economy but that seems a bit far fetched. Possibly there was some global boom that impacted both the US and UK. Shock A does correspond with a massive spike in RGDP growth (see below). Shock D corresponds to RGDP growth that is high as well ‒ at least high relative to the post-financial crisis norm. If anyone out there has some theories as to what could be happening, let me know in comments.
I don't think I went through the "quantity theory of labor and capital" for the UK before. The model is basically a modification of the Solow model and gives us functions NGDP = NGDP(L, K) and CPI = CPI(L, K) = dNGDP(L, K)/dL. Here are the 4-parameter (+ 1 normalization factor that drops out of the rates shown below) fits to the data:
I show year-over-year growth rates in order to cut down the seasonal noise. The NGDP growth is a bit low and CPI inflation is a bit high meaning the CPI-deflated RGDP model (blue in the third graph) is significantly below the CPI-deflated RGDP data (yellow). The year of year change of the RGDP data reported in FRED is shown in green. Not bad for almost 50 years of data for two (independent) time series and 4 parameters.
III. The two together
Additionally, we can see the rough correspondence between the RGDP growth rate (green data from above) and the employment shocks from the dynamic equilibrium model in the graph below. This isn't necessarily groundbreaking research as it is basically a confirmation of Okun's law.
We end up with a fairly simple information equilibrium model that captures most of the empirical data that has 5 parameters:
NGDP ⇄ K
U ⇄ CLF
where U is the total unemployed, CLF is the civilian labor force, L is the number of people employed, and K is the nominal capital stock (from here). The five parameters are the IT indices of the three relationships (α, β, and k, respectively with the former two being the Solow exponents) along with two normalization factors (L₀, K₀).
Shock B closely corresponds with the 1997 general election, which was when "New Labour" gained power, so maybe that has something to do with it. 'A' isn't as close, but Thatcher left office in 1990, only about a year after 1988.8. As for C and D, I have no idea.
ReplyDeleteI had thought of the government transitions, but couldn't point to policy changes that were significant enough. However that might be my lack of knowledge of UK policy.
DeleteB does correspond to setting up the MPC at the BoE ...
I did a bit of research into unemployment and economic growth a while ago. The total GDP figures throw up the odd anomaly as you have shown. When looking at data from the 1960s-70s, it's best to look at the performance of manufacturing, as that was much more of a growth engine back then, and then for later decades (1980s onwards), I found that it's quite useful to look at consumer spending, when it comes to rapid changes in joblessness.
ReplyDeleteThe sharp fall in joblessness during the late 80s was in large part due to low productivity growth and the so-called "Lawson Boom". The Tories over-stimulated the economy, and assumed they'd raised the potential growth rate of the economy. However, productivity weakened in 1988-89. Inflation rose, so they hiked rates and this led to the 1990s recession.
The 1997 drop in unemployment appears to have been linked to a rise in consumer spending in 1996-97. At least that's what I assumed. St Louis Fred is a good tool to chart the similarities of employment growth and consumer spending.
I found that the rise in unemployment (Shock C) wasn't necessarily due to weak economic growth. It was more likely due to an increase in the number of people of working age who were briefly unable to find work. This could be to do with immigration or something else.
As for Shock D, things are a little complicated. The financial crash probably lowered trend growth from about 2.7% to about 2%. It's effectively punched a permanent hole in the economy's productive capacity. By 2013-14, inflation was falling, forward guidance was introduced and the economy was improving but it was only when growth crossed the 2% level that we started to see unemployment falling substantially.
DeleteThe 3.1% growth seen in 2014 would have probably meant fairly steady unemployment in the pre-2008 era. However, because of weaker trend growth, that's probably enough to constitute a boom nowadays, hence the sharp fall in unemployment in 2014. Like to know what you think.
Thanks for the information.
DeleteI did notice that the post-crisis growth in the UK has been associated with falling productivity in the various analysis -- effectively a lot more people employed producing only a little more output.
When you mentioned immigration it made me think that adding several countries to the EU in 2004 may be behind the rise in unemployment for shock C. Poland (for example) had a fairly high unemployment rate at the time and the 2004 enlargement might have lead to some "averaging effect" across the labor market.
This is an interesting post. From my British perspective, I have a few comments.
ReplyDeleteFirst, although you suggest global forces are mostly responsible for major shocks, the two largest unemployment peaks shown in your charts coincided with UK-specific events. The peak in the 1980s coincided with Margaret Thatcher’s monetarist experiment, while the peak in the early 1990s coincided with the UK’s membership of the ERM. This was the predecessor of the Euro. The UK pegged its currency at too high a rate and was forced to withdraw from this arrangement by the markets in September 1992.
https://en.wikipedia.org/wiki/European_Exchange_Rate_Mechanism
https://en.wikipedia.org/wiki/Black_Wednesday
These two events were amongst the most significant economic events of the second half of the 20th Century for the UK, as we are still dealing with their repercussions. The Scots have never forgiven Thatcher for monetarism, and this is still at the heart of the continued upsurge in support for Scottish independence. The ERM was the beginning of the end for Britain in the EU. It’s not an accident that the people who are old enough to have been adults in the early 1990s are noticeably more Euro sceptic than the younger segment of the population. The ERM – complete with high unemployment and interest rates which peaked at 15% - saved the UK from the Euro. It’s also why British economists, such as Wynne Godley, were at the forefront of criticism of the proposals for the Euro. Here is an article by him written the month after the UK left the ERM.
https://www.lrb.co.uk/v14/n19/wynne-godley/maastricht-and-all-that
Second, you don’t suggest any shock around April 1999. This was when the UK introduced the national minimum wage, so you seem to be saying that the minimum wage didn’t cause any noticeable shock to the unemployment figures.
https://en.wikipedia.org/wiki/National_Minimum_Wage_Act_1998
More generally, I am always surprised that the macroeconomics profession does not seem to have a master database of possible causes of economic shocks, so that an analyst in your position can easily identify what causes might have led to a specific effect in charts such as the ones you show here.
Hi Jamie, I was hoping you'd share your perspective.
DeleteLet me take some time with this.