dailynewsnblog
Transcript: Guns and Butter
Tim Phillips talks to Christoph Trebesch
Christoph Trebesch
The long-run legacy of these choices is indeed higher taxation.
Tim Phillips
At a time when Europe’s NATO members have pledged 3.5% of GDP to rearmament, What does history tell us about the hard choices they might have to make to fund that commitment? My guest today has analysed 150 years of military spending to discover how governments really make the choice between guns and butter. Welcome to Vox Talks Economics. I’m Tim Phillips. Politicians through the years have often talked about the choice of guns and butter. This metaphor captures the choice between social spending and spending on weapons. Christoph Trebesch of the Kiel Institute was one of the researchers who has run the numbers on how this trade-off has historically turned out. He joins me now to talk about this painstaking work. Christoph, welcome back. Glad to see you again.
Christoph Trebesch
Thank you very much.
Tim Phillips
What do we mean precisely by this metaphor, guns and butter, in this context?
Christoph Trebesch
There’s this classic view that there’s a trade-off between spending on the military and spending on other civilian purposes, in particular social purposes, so health, pensions, education, etc. So there’s this idea that the military spending crowds out social spending. That’s what the core of this guns versus butter metaphor typically means.
Tim Phillips
One of the reasons why we think about it is, when you look at how much military spending costs, sometimes you think, ‘crikey, that’s a lot’. How large are the sums committed to either fighting wars or preparing for them, compared to the other government spending that we’re worried about this crowding out?
Christoph Trebesch
So it really depends on whether you look at today where social spending is 30-40% of GDP in advanced countries compared to just 2-3% for the military. So today the military is a small part of of the total budget. In history, it was a larger part of the budget because the welfare state wasn’t as developed. So it really depends on period. What is clear is that during wars, military spending goes through the roof. Countries were spending 50, 60, 70% of GDP on the military. And this is what also happens already today in Ukraine. In wars, military spending is absolutely inconceivable from today’s perspective. In peacetime, the numbers are smaller, but yet they are meaningful. They can be very large economically, even in peacetime.
Tim Phillips
In Europe, we’re all talking about this at the moment, partly because of Ukraine, partly because of the controversy about what countries should be committing to NATO. But I imagine that this argument about where this trade-off should be made, I imagine it’s not a new one. How long have people been arguing about this?
Christoph Trebesch
That’s actually fascinating about the project. You go back to parliamentary debates in the 19th century, and even then there was a discussion on cuts in social or other productive spending purposes of governments. So this movie has been there many, many times.
Tim Phillips
What questions did you try to find out?
Christoph Trebesch
So, we are most curious to understand the fiscal consequences of this large military buildup that we are currently seeing. So, we try to learn from history because we haven’t had this kind of military buildups for decades. So, we have to go back to the 50s. to find a similar period in which as many countries as today in NATO are building up simultaneously at this large scale. So, their history comes in as very useful because just using the last 30 to 40 years, well, you can look to some developing countries, but generally there was this tendency of just not investing as much in the military. So, we asked ourselves, what’s the fiscal consequences? How were these trade-offs addressed before? And we looked at debt financing. We looked at tax financing. And, and this is quite new, we also looked at within budget reprioritization, right? So, do we see this cut in social spending, this cut in butter in periods of rearmament, in periods of war? Surprisingly, despite this notion being around for decades, there is no in-depth work across countries and across time that has addressed that question, right? So, there’s excellent literature on individual countries, typically looking at the debt versus tax question. And then there’s work looking at, you know, social spending versus military, but bringing it all together and putting it in a large historical perspective, that’s really what this paper is trying to do.
Intermission
If Europe decides to re-arm, it has the industrial base. But in our episode called Can Europe Defend Itself? that aired on the 14th of January this year, Moritz Schullerick also at the Kiel Institute, warns us that it isn’t converting that capacity into credible deterrence. Find out what European rearmament could mean in practice by listening to this episode when you follow us, wherever you get your podcasts.
Tim Phillips
So, Christoph, one of the first things I noticed is that you don’t only investigate wartime spending. Why not?
Christoph Trebesch
Because the situation we currently face in Europe is one in which, you know, Germany is not at war and neither is France or the UK. And yet we have decided to double or triple our military outlays. Understanding the fiscal and economic consequences of this decision of quickly increasing one subset of the budget on such a scale, to us was a first-order question given policy today. And nobody has really looked into these peacetime military build-ups. So, we thought that is an important addition to knowledge to the policy debate. There was systematic work on war spending. But we try to push that as well and drill deeper and understand in more granularity what was really going on in the financing of war. So, we try to push the research frontier on both sides, both on digging deeper on wars, what was going on, and understanding for the first time systematically about peacetime rearmament.
Tim Phillips
How in that case do you define a boom, unfortunate name in this context, in military spending?
Christoph Trebesch
We use the credit booms literature. This was our inspiration. You know, the last big crisis we had was financial. And there was all this research that trying to understand the consequences of these credit booms. So, we use this kind of methodology and identify booms by strong increases in military spending to GDP. Specifically, if a country has a 6.5% increase in military spending to GDP over two consecutive years, that is considered a boom onset according to our algorithm. And then the boom ends once the growth rate goes below zero. So, when basically there’s no more growth in military spending. And that’s our algorithm. And then we just start using historical narratives. So, case by case, we really drill in to make sure we are not misclassifying because the algorithm picked up some GDP decline or something like that. And then we end up with a set of 116 cases that we know from both the algorithm and the data and from the historical record that this was really a military buildup.
Tim Phillips
This is quite data intensive work already. We’ve only just started talking about it. Where do you find the data you need for this?
Christoph Trebesch
This was indeed quite an adventure. It’s been a four-year process.
Tim Phillips
Four years!
Christoph Trebesch
Yeah. So, the beauty of working with government budgetary data is that these governments have to make sure these data are correct in democracies, because there’s parliamentary oversight: this is actually a law enacted in parliament. So, there’s a lot of scrutiny. compared to other fields like I’ve worked on capital flows. It’s a mess because there’s no legal obligation to do this right. Whereas for budgets, you know, there’s a lot of people watching closely. But nobody had really unearthed all this raw primary… source material, looking at the actual budgetary archives of countries. People have used secondary sources which then mixes up planned versus realized expenditures, has a lot of errors, ignores off-budget items which often play a major role in wars. We dug and we looked at these primary sources and they’re beautiful. And then you put them all together and for the first time can create this long run history of what governments actually spend on, which is by itself a very fascinating aspect of this project.
Tim Phillips
There must be a lot of manual work in getting this into shape. You can’t do this all with an algorithm, especially if you’re going back 150 years.
Christoph Trebesch
It’s a mix of manual work. So, we had a lot of RAs. I mean, Johannes Marzian was really in the driver’s seat on this, my co-author. We had a lot of RAs, but we also leveraged AI and used tools to extract automatically. So, it’s a mix. We did benefit a lot from the technological advances that generally our field is benefiting from. Would have been much harder without that. You need somebody who really becomes an expert on budgets. And Johannes certainly is that.
Tim Phillips
And from this, you create this new data source. What’s in it? How comprehensive is it?
Christoph Trebesch
So, we cover 20 countries back to 1870 with some gaps for some countries like the Soviet Union at some point. Most advanced countries, annual, so for every year, you know, there’s hundreds of pages of these budgetary documents. The main addition or the main added value, besides having much cleaner series of these budgets, it’s really that we have a breakdown by purpose. So, we have interior, you know, police and justice system, we have for the foreign office, we have of course the military and we have the social or infrastructure spending, right? So, you really know what the government is spending on. That’s the key that this breakdown by ministry or by purpose is really the main novelty to have that back to the 1870s. That just didn’t exist.
Tim Phillips
A lot of what we know about this comes from secondary sources. It’s already been interpreted when you go back to the source Are you finding surprising information, things that we really didn’t know before?
Christoph Trebesch
We realize just how badly we had been measuring before. So, let me give you an example. Switzerland, there’s the official Swiss National Archives, and this is a very well-known data source of the correlates of war, tracking military spending. Now, if you look at these series, then, Switzerland, during the World Wars, was a very peaceful nation, relied on their neutrality and didn’t invest much in the military. Because according to the data, it’s basically flat, it’s 2% of GDP that they were spending. But once you drill deeper, you look into the budget documents and really scrutinize it, you realize that there was this huge off-budget… item in place during the wars and then spending is actually 10%. The military spending actually goes up from 2% to 10% once you count this budget. So it’s five times. In other words, Swiss massively re-armed during the world wars. And this is very much in line with what historians and qualitative sources say. So, if you trust previous data sources, you get a distorted picture for many of these countries. Plus, many of the existing data sources used planned expenditures. But you can imagine in the worst, realized expenditures are often much, much higher than planned. So, that’s another added value of the database. We realized how big the actual spending is, and we can trace it.
Tim Phillips
This data must also be quite messy, unavoidably messy, because there are a lot of things going on during wars. Can you make a causal claim when you’re talking about the impact of military spending when the conflict has changed so many other things as well?
Christoph Trebesch
This is of course a major challenge. We tried our best, let me answer it this way. So, what we did are two things. We first identify external shocks, geopolitical shocks, right? For example, France in the mid 1930s was faced with a very aggressively rearming Nazi Germany. A leftist government came in power and they wanted to spend on social issues, but given the threat from Nazi Germany, they nevertheless went into a military spending spree. And these kind of settings where you have an exogenous trigger that has nothing to do with the domestic political economy, how the economy in France is going at the time, whether there was a recession or not, These cases of exogenous sudden shocks, if you want, they are coded by us as exogenous peace time, with their spending booms that are triggered by outside events, outside of the control of the country. That’s one. And we do the same for wars, right? This is a surprising war, this is an exogenous war. So, when Germany attacks France, then that is an exogenous imposed war for France, but it’s not one for Germany. Right? For Germany, it wouldn’t be exogenous because they started it. So, that’s our classification of exogenous events. We then run regressions, local projections, also some simulations, to understand the long-run path of fiscal variables of taxation after these exogenous shocks, according to the classification. Now, of course, there’s limitations to that, so what we were also doing, once you see the aggregate result, we were puzzled, by just how persistent these tax increases are, and the tax rates are. So, what we did is go case by case, and really try to find out which taxes were increased, which taxes were introduced, and whether the tax increases and directions were taken back after the war. So, we do this, basically, many tax histories around military booms, and really go into the parliamentary … we go into the archives, we look at historical work and really try to understand what we observe in the estimates. Can this be observed on the ground? Is it real? Is Parliament talking about it? Are people complaining that the taxes are still high after the war, etc., etc.? And there, of course, also AI helped a lot in finding the narratives, but also, you know, basically, you find the events. It’s amazing, right? The AI helps you to source, you know, some tax legislation in Norway of 1875, you know, And, then the tax legislation has in there, well, this tax was introduced because of the war or because of the threat of Sweden. And then you can really find over time when they abolish the taxes again. So, you get to the actual raw sources using AI. We don’t rely on what the AI tells us. We can really go to the original sources, and the AI helps us to identify those sources. That, again, was an amazing journey, just learning so much about these events.
Tim Phillips
But equally, if you just rely on the data, it might send you in the wrong direction for this. So the combination of the two is essential.
Christoph Trebesch
Yes, the quantitative approach and the qualitative both tell a similar story. And this lets us conclude that there is something here and that there seems to be a large legacy of these events.
Tim Phillips
So, for anyone who started off thinking, ‘that took them four years?’, after that last description, they’re thinking, ‘that only took them four years?’. I’m now going to distill your four year research into a few very simple questions, Christoph. These booms in military spending, are they financed by debt or are they financed by taxation?
Christoph Trebesch
Wars, which are temporary in nature, typically, are largely financed by debt. That is something we knew already. We only confirm it in a much larger data set, more systematically. Peacetime booms, you know, those outside of war, you’re not fighting on your soil. Peacetime booms are financed through a more balanced mix of taxes and debt. And peacetime booms tend to be somewhat more permanent, take longer. And that, again, is very much in line with theory, which predicts that sudden and temporary expenditure booms should be financed by debt because you can smooth taxation over time. Whereas these lengthier, deterrent investments, as we are currently observing out there, they are financed with both debt and taxes. There’s a mix and there’s quite a bit of heterogeneity across cases.
Tim Phillips
Is this choice between guns and butter a real one? Do military booms result in cuts to social programs?
Christoph Trebesch
So, this was one of the bigger surprises of the paper. One of the unexpected results is that we do not find many cases in which social spending is cut. It’s just very rare to find that the military crowds out social spending on a large scale. And this is very much at odds with our prior and very much at odds with this general belief. that we do not find many cases Social spending might not grow as fast as it maybe used to, but it keeps up with inflation. That’s what we find. So there’s less of this reprioritisation than we thought. So that’s why the paper is titled Guns and Butter, because that’s what apparently governments mostly choose. They chose both. They choose to increase social spending or at least keep it constant at the inflation growth rate when they rearm.
Tim Phillips
In that case… What happens to taxes?
Christoph Trebesch
The long-run legacy of these choices of going up with military spending and social spending is indeed higher taxation. You see an expansion of total expenditures and you see an expansion of the fiscal state and the taxation state. So, you see increases in the top income rate of 10, 15 percentage points, quite massive in the decade following a military boom. And you see significantly higher tax revenues. Most of these increases come from broad-based taxes. We’re talking about income taxes, value-added taxes. These kind of broad-based taxes are really what seems to finance military booms.
Tim Phillips
And so this brings us back to the policy choices that politicians are making in 2026. What does this tell us about how they are going to pay for re-armament, and I guess also how they’re going to sell that to the electorate?
Christoph Trebesch
We finished the first draft of this paper in 2024, something like that. I thought it would never be possible that we would see the same pattern again. Taxation is already so high. Debt is already so high. And yet what we are seeing is is that in most countries [they] are financing this military buildup with more debt and now including in Germany there’s talk of higher taxes in a way, our sample prediction is turning out shockingly in line with what I would never have expected because I got a lot of pushback in seminars. Well, this time is so much different. We have these large welfare states. Surprisingly, it seems that at least in the short term, the patterns seem to repeat themselves. More debt and more taxes are now, with some delay, you know, three, four years after the boom started, are now apparently in the cards. So, let’s see how things develop. But broad-based butter cuts are not being discussed. A little bit in Germany now on health, but overall the first reaction was debt and now taxes.
Tim Phillips
This time it turns out is not different. Christoph, thank you very much for talking about your research. It’s fascinating.
Christoph Trebesch
Thank you very much.
Tim Phillips
You should read this. It is very on the nose for 2026. The title ‘Guns and Butter, the Fiscal Consequences of Rearmament and War’. It is Discussion Paper 21193. The two authors, Johannes Marzian and Christoph Trebesch.
Outro
VoxTalks Economics is a Talk Normal production. The assistant producer is Megan Bieber, and our editor is Andrei Zagarian. Next week on VoxTalks Economics, redefining the monetary standard.
