Baxter & King 1993: Why government spending isn't just moving money around

This is the first in a series of posts called "Papers You Should Know". I believe one of the essential roles of econ blogs should be to bring research into the public consciousness. There's a ton of really interesting papers out there that get mooted in academic circles, or taught in grad classes, but whose insights never really make it into the wider discussion. That's a shame. It means that most intelligent, educated non-economists encounter a hundred worthless American Enterprise Institute hackonomics propaganda pieces for every one serious piece of scholarly research. Which, I find, often leaves people with the vague notion that academic economists spend 90% of their time glorifying the frictionless perfection of the free market, and the other 10% collecting "consulting" fees for big banks.

It isn't so. There is absolutely tons of research out there that looks at questions of how markets break down, and how complex "frictions" make interesting stuff happen. Today I'm going to look at one such paper: Marianne Baxter and Robert G. King's "Fiscal Policy in General Equilibrium" (American Economic Review, 1993).

First, for some background. One of the biggest questions for macroeconomists right now is whether government spending can boost the economy. You often hear conservative-minded economists saying that no, government spending is just "moving money around" without creating any wealth. For example, here is John Cochrane, of the University of Chicago, from 2009:

[M]oney [spent by the government] has to come from somewhere. If the government borrows a dollar from you, that is a dollar that you do not spend, or that you do not lend to a company to spend on new investment. Every dollar of increased government spending must correspond to one less dollar of private spending. Jobs created by stimulus spending are offset by jobs lost from the decline in private spending.

Where this idea really comes from is an academic paper - in fact, one of the most famous econ papers ever written. In "Are Government Bonds Net Wealth?" (Journal of Political Economy, 1974), Robert Barro (currently of Harvard) showed that under certain conditions, the timing of government spending doesn't matter for the economy. This idea, which is called "Ricardian Equivalence", is exactly what John Cochrane was talking about. In this model, government can only move economic activity around, not create it.

When I read this famous paper, I immediately found it fishy. One of the model's (unstated) assumptions is that government spending consists entirely of transfer payments - i.e., taking money from one person and handing it to another. In other words, Barro assumes that government is fundamentally different from, say, a corporation; while a corporation can invest money today to create new wealth tomorrow, government can only shuffle money around. There is no "government capital" that we can invest in today that will create new wealth down the road.

To me, this felt a bit like assuming the conclusion. If you just assume government doesn't produce anything new, of course it's going to be hard to get government spending to boost the economy! If you think that public goods can boost the economy, then everything changes. I pointed this out in a blog post back in March, and even considered writing a paper about it. Then I found out that someone had already written that paper. And those someones were Marianne Baxter and Robert G. King, in 1993.

Only scooped by 18 years. Not bad, eh?

What Baxter and King (here's the link again) do is to take your most basic neoclassical business cycle model - the RBC model - and simply add government capital. What is government capital? Well, it's any kind of capital that the private sector can't or won't build (or can't or won't build enough of). In other words, government capital is a nonrival production input, or "public good." For the math on how those work, see here. Basically, these are things like roads, electrical grids, airports, and other infrastructure. Schools and research centers and courts and police could also count.

So Baxter & King put public goods into the production function. Instead of only including private capital and labor, GDP now depends on government capital as well. Here's the equation, for those of you who like equations:

Everything else is just pure RBC - no frictions, no sticky prices, no sticky wages, no financial sector, no nuthin'. Exactly the way Ed Prescott would like it, except for that little KG thing there on the end.

And of course, what happens? "Ricardian equivalence" goes straight out the window! If the government can invest in useful projects just like a firm, then the timing of government spending matters a lot, just like the timing of private investment. Baxter & King find that, in their model, government spending has a huge "multiplier", even with none of the Keynesian stuff like sticky wages.

But this is hardly surprising. Like I said, Barro starts with the assumption that government spending is 100% transfers. It's almost painfully obvious that if you allow for government to actually build useful things, you're going to get a very different result. In fact, while giving full props to Baxter & King, I'm a little surprised it took until 1993 for people to do this.

So, do the data support Baxter-King? Well, the short answer is that I need to look into this a lot more. But here is an interesting paper by Michigan's own Rudi Bachmann, along with Eric Sims of Notre Dame, that shows that when the government invests more money, especially during downturns, it raises business confidence. The last line from the abstract basically says it all:

In particular, spending shocks during downturns predict future productivity improvements through a persistent increase in government investment relative to consumption, which is in turn reflected in higher measured confidence.

Which is exactly what you'd expect from Baxter-King.

Now let's return to the present debate over government spending. Most of that debate revolves around things like liquidity traps, sticky wages, business confidence, and so on. But if we live in a Baxter-King world (or a Bachmann-Sims world), the case for more spending is actually a lot simpler. We desperately need to repair our country's infrastructure. Only government will do that. Interest rates are historically low, meaning that now is the perfect time to borrow money to rebuild the roads. Doing so would raise employment today, but because roads are necessary for tomorrow's businesses to thrive, it would also raise GDP in the future. It's not just a slam dunk, it's a free lunch!

There is, of course, a caveat here. Government capital has its limits. As Japan proved when it concreted over its rivers and built bridges to nowhere in the 1990s, you can reach a point where more infrastructure is just inefficient (something that Baxter & King, notably, do not allow for in their model). But I would argue that, given the dilapidated, crumbling nature of America's roads and bridges, we are not anywhere near that point. 

So to sum up: Baxter & King (1993) show that it's very easy to get government spending to matter in a big way. All you have to do is assume that public goods exist. Every time you hear someone say that "the money [for stimulus] has to come from somewhere!", just tell them that the money will come from the future wealth that private businesses will create once they can ship goods to each other over our shiny, new, government-built roads. Or just say "Baxter & King, 1993".