The U.K. Office of Budget Responsibility (OBR) published its “Fiscal Risks and Sustainability” report (link) and it contains the eye-catching chart above. I will draw your attention to the top line which represents a “worse case” projection of the debt/GDP ratio marching merrily off to 1,000%. I do not expect to be able to pay off on any bets in 2076, but I find it safe to say that the debt/GDP ratio will not do that. Even the low end projection is implausible.
I will first explain why the high projection is nonsensical, which then leads to a discussion why any methodology that produces such a scenario is unsound.
Why A 1,000% Debt/GDP Ratio Is Effectively Impossible
A country could do something remarkably stupid and end up with a debt/GDP ratio of 1000%. A very small country ruled by a wealthy monarch that has large non-domestic holdings might put themselves in such a position (most likely in a Europa Universalis V run). However, it is not going to happen for the sorts of governments we see in the developed countries outside of some national disaster scenario (which is not what the projections represent).
To see why, we need to decide what is a low-end estimate for nominal GDP growth. Assuming that the U.K. is somewhere near its 2% inflation target (and eliding the difference between consumer price inflation and the GDP deflator growth), 1% annual nominal GDP growth seems like a lower bound for the average. This allows for near-stagnant real GDP growth and undershoots of the inflation target. Even though the workforce is projected to shrink, there should still be some improvements to raw labour productivity (output per labour hour) due to technological improvements and capital deepening. (Even if you are pessimistic about the long-term outlook, people are still likely to be working and producing something, even if “standards of living” are dropping — the economic activity will still raise measured GDP.)
At 1% nominal GDP growth, the steady state deficit for a 1000% debt/GDP ratio is 10% of GDP. That is, since the debt level is 10 times the level of GDP, to keep the debt/GDP ratio constant, the level of the debt has to increase 10 times as fast as the level of GDP.
In summary, we have to believe that the government would continuously run a deficit of 10% of GDP yet the economy is barely growing in nominal terms.
One can try to point to Japan as an example of that sort of situation. You did get a combination of slow nominal GDP growth and large deficits. However, the net debt GDP ratio capped out at 160% (using the IMF annual figures (link). Japanese gross debt figures are over 200%, but governments lending to themselves is an activity that has no effect on the macroeconomy. (Although it would be possible to reach a 1000% debt/GDP ratio by making a sufficiently large loan to yourself, that is not useful information for real-world fiscal policy.)
Increasing nominal GDP growth rates makes the 1,000% projection look even sillier. At 5% nominal GDP growth, the government deficit needs to be 50% of GDP. There is a reason why net debt/GDP ratios tend to cap out between 100-200% with nominal GDP growth rates running at 3%-5%; GDP growth will cut away at the ratio as soon as the economy is moving away from recession.
How Did the OBR Come Up With Dubious Projections?
The Office of Budget Responsibility followed a complex analysis path that is blessed by neoclassical academics. The idea is that the real side of the economy follows fundamental forces over the long term (productivity, labour force growth) that we allegedly can project independently of the business cycle. They then attempt to extrapolate tax revenues and expenditures based on current policy settings and the extrapolated real values. (For example, oldsters consume more health care expenditures, so they can extrapolate future health spending based on demographic projections).
Although this is the “serious” and “sophisticated” way to do this, it faces a fundamental problem: tax revenues and government expenditures are big numbers, if we extrapolate growth rates for them, the difference (the fiscal deficit) is a big number that will get extremely large if the revenues and expenditures do not have the exact same growth rate.
In the real world or in more sensible economic models, the economy reacts to a fiscal deficit. If spending grows faster than revenue, the deficit add fiscal stimulus that causes growth acceleration. Greater nominal growth reduces the need for welfare spending, and greater nominal incomes means that the tax take should increase. (This used to be called “automatic stabilisers,” but the neoliberal turn and neoclassical theory discounted the importance of them. Instead, neoclassical models feature economies that are stabilised by central banks manipulating expectations fairies.)
Any model that predicts that revenues and expenditures will grow feature markedly different average growth rates for 50 years is worthless since it ignores the interactions within the system.
Real Worry — Inflation
For a country that is borrowing in its own currency and does not allow incompetents to deliberately sabotage their own bond market (see everyone involved in l’affaire Truss-Kwarteng), “unsustainable” fiscal settings will sooner or later cause an inflationary accident. According to neoclassical theory, the expectations fairy will cause fiscal policy problems in 50 years manifest in an inflationary explosion right now. It seems more plausible that markets are forward looking, but not that forward looking.
It is entirely plausible that if current policy settings were unchanged, the United Kingdom would run unto problems some time. However, it is completely unreasonable to blow up your economy right now based on a projected problem fifteen years out. I do not advocate attempting to fine-tune the business cycle with tax rate changes, but it is entirely sensible to raise taxes every so often if the economy is leaning towards overheating or new programmes are being rolled out. (I have not followed the U.K. economy in enough detail to have a strong opinion on the current stance of fiscal policy.)
What Can We Do?
Our ability to model the economy quantitatively just 1-2 years out is poor, attempting to do so on a multi-decade horizon is pure wishful thinking. Realistically, a 4- to 10-year planning horizon is the most useful exercise, and even those projections will always crash into reality. Those shorter projection horizons will still have dubious extrapolations, but they will have less time to spin off to ridiculous numbers. Their value is offering some guidance on the medium-term stance to policy. (MMT purists might object to deficit-based analysis, but to a certain extent, they do tell us about the overall policy stance, although we cannot attach too much value to particular levels. See discussion in next section.)
Subsets of government spending can be approached on longer horizons. Defence programmes have long lifetimes. School and hospitals are driven by demographics which you might be able to project. Another contentious area is state pensions. The issue with state pensions is more the perceived fairness of the system. To bolster the political strength of the programme, you need people to believe that it is “their” money that they are getting back, so you need to make the system appear to be actuarial sound. Even though these long-term planning exercises might be useful for those spending areas, this analysis is completely decoupled from the rest of the economy, so we have to be cautious regarding how meaningful they are from a macroeconomic perspective.
In any event, it does not make sense to tighten fiscal policy now because you are worried about inflation twenty years from now. But one might say, we could lower debt levels now — which is why “serious” economists all hopped onto the austerity train in the 2010s. The problem is that “austerity” is invariably “cut spending on social programmes that benefit the poorer segments of the population,” which has the effect of undercutting growth rates. By crushing growth, they made debt/GDP ratios increase. If you want to decrease debt ratios, you need to maximise revenues without hitting growth — which you can do via hammering the rich with tax hikes. (More on this point below.)
Aside: MMT and Deficits
In this article, I used deficits as a shorthand for the stance of fiscal policy. There are a few qualifiers that should be noted.
-
There can be wacky things going on with government accounting that introduces a gap between the fiscal deficit and the government’s effect on income streams in the economy. These are normally going to be temporary, and not worth complicating the text to discuss.
-
Other balances in the economy affect the “steady state” fiscal deficit. For example, countries running persistent current account deficits (typical for “anglo” economies in recent decades), the government typically needs to run a deficit to offset the drain of income to the foreign sector. Conversely, current account surplus countries might need to run fiscal surpluses to counter-act their exuberant export sectors. (I believe Australia was in an interesting position with regards to those statements, but I believe a lot of the foreign accounts were the result of transactions by large multinationals). In addition to the external sector, developed economies feature ageing populations that are saving for retirement — creating a drag on growth.
-
The composition of spending matters. It is possible to send out a lot of money yet have little effect on growth, as the money is saved. (Tax cuts to rich people being a prime example for the post-1980 period.)
The fundamental problem with neoclassical treatment of fiscal policy is that they force the economy to be the result of an optimisation problem of households. This does not leave a whole lot of room for other actors. Central banks feature because of their reaction functions for setting interest rates, but fiscal policy essentially disappears. All the government does is set exogenous tax and spend trajectories, and then they effectively disappear from the model if those trajectories are “sustainable.” The behaviour of households is furthermore probably too “rational” and monolithic.
Simpler models like the Post-Keynesian stock-flow consistent models (SFC models — see my introductory book on them) feature a household sector that follows simpler rules. The consumption function can be interpreted as households having a target stock of wealth (which includes money and government bonds). What we see is that if the household sector “wants” to increase its stock of wealth, it will slow the economy until that target is reached. Automatic stabilisers will result in wider deficits — which provide the assets needed by households to hit their target wealth levels.
The data are quite clear about the post-1980 trends: disinflation coincided with a massive growth in the wealth-to-income ratio of the household sector. This is partly due to increasing inequality, and partly due to rise of ageing middle classes that are saving for their retirements. Although equity and real estate holdings represent most of the increase in household wealth, government debt holdings also feature.
If you want to lower government debt-to-GDP ratios, you need to figure out a way to reduce household assets. Have fun selling that programme!
Concluding Remarks
The output of economic institutions has an inherent weakness in practice — there is a desired end goal, and then the economists work backwards to determine what analysis gets you to that target. This is almost certainly what is happening here — the analysis fits the desired “we are serious about debt levels” political framing, and assumes that neoclassical modelling of fiscal policy is correct. The complexity of the analysis provides a distraction from the obvious problems of the projection outputs. For people whose full-time job is to write serious reports about fiscal sustainability, admitting that multi-decade forecasts is an impossible exercise is a career-limiting move.
