Why Assumptions Matter So Much
Veronique de Rugy and Doug Holtz-Eakin emphasize the relationship between assumptions and CBO projections
Many sciences - economics especially - deal heavily in projecting into the future. What makes such projections science, rather than art or augury, is the combination of methodological rigor and empirical scrupulousness. Yet, unlike “harder,” experimental sciences, macroeconomics must accept a lack of confidence in its results if it is to say anything useful.
If your data are compromised or do not represent what you actually want to measure, it throws off your results. Garbage in, garbage out. If the model that you’ve constructed, into which you are plugging the data, is inappropriate or poorly-reasoned, the same thing happens. If your baseline assumptions - parameters in the mathematical sense - are wrong, ditto. Moreover, you could make assumptions appropriate at the time of modelling, but turn out to change drastically over the period of projection. There are manifold sources of uncertainty - a weakest-link sort of situation. The result of the famous Simon-Ehrlich wager (or just good ol’ Malthusianism) is a good illustration of how this often plays out. A course or book covering introductory empirical social science, or even just statistics, will give you an even better idea of the minefield modellers face.
Making assumptions under different plausible scenarios is essentially Step 1, and arguably the most influential on the ultimate results. When CBO makes projections and scores proposed legislation for Congress, it is keenly aware of how much this matters. Therefore, they generally include different possible future scenarios, showing how exogenous factors can change the outcomes of modelled policy.
Our own Veronique de Rugy addressed this issue in her recent column for Creators Syndicate. She takes aim at her bête noire: fiscally irresponsible politicians.
So much can and likely will happen to make projections moot and our fiscal outlook much grimmer. Unforeseen events, economic changes and policy decisions render them less accurate over time. The CBO knows this and recently released alternative scenarios based on different sets of assumptions, and it doesn't look good. It remains a wonder that more politicians, now given a more realistic range of possibilities, aren't behaving like it.
Current assumptions are relatively rosy, yet we are still on track to breach a 100% debt-GDP ratio within 10 years, and 160% within 30. De Rugy writes:
Unfortunately, if any of the assumptions underlying these projections change again, things will get a lot worse.
…
For instance, the CBO highlights that if the labor force grows annually by just 0.1 fewer percentage points than originally projected — even if the unemployment rate stays the same — slower economic growth will lead to a deficit $142 billion larger than baseline projections between 2025 and 2034. A similarly small slowdown in the productivity rate would lead to an added deficit of $304 billion over that period.
The same applies to interest rates, which although still low by historical standards, have been trending upwards:
Even a minuscule 0.1-point rise above the baseline would produce an additional $324 billion on the deficit over the 2025-2034 period.
And inflation…
To be precise, an increase in overall prices of just 0.1 points over the CBO baseline would result in higher interest rates and a deficit of $263 billion more than projected.
Now, imagine all these variations from the current projections happening simultaneously. It's a real possibility. The deficit hike would be enormous, which could then trigger even more inflation and higher interest rates.
She closes by emphasizing the need for Congress to take the power of these assumptions seriously if we are to avoid a fiscal death spiral.
In a similar vein, former CBO Director Douglas Holtz-Eakin wrote a blog post for the American Action Forum, covering the imbalances between the statutory assumptions on the tax vs. spending sides of the equation.
One way to think of this is that taxes follow current law in the baseline, while spending follows current policy. However you label it, the two sides of the budget are treated differently, with the result that Congress must pay for extending tax laws but faces no deficit consequences for extending spending programs.
I would recommend reading the post in detail, as well as the linked CBO explanation, but he offers some specific examples of how this is the case:
… because current law is for TCJA to largely sunset at the end of 2025, the baseline shows a large tax increase beginning in 2026. Any law to extend TCJA would reduce taxes compared to this baseline, increasing the deficit.
…In contrast, spending is governed by the “$50 million rule,” which states that any program with more than $50 million in outlays – virtually every program – is extended in the baseline automatically even if the program is no longer authorized. Notice that if a law subsequently reauthorizes the program, the comparison with the baseline would show no increase in spending or the deficit.
Holtz-Eakin concludes by advocating that this imbalance be corrected, either by treating taxes like spending (i.e. extending current-policy assumptions), or ideally, vice versa (extending current-law assumptions):
This would mean every dime of annual appropriations would raise the deficit – because it does. And every reauthorized mandatory spending program would raise the deficit – because it does. And extending tax cuts would raise the deficit – because it does.