Last week, Macroeconomic Advisors (MA) ran a blog post that expressed reservations about the estimate of the growth-impact of increased public investment that accompanied EPI’s analysis of the People’s Budget – a plan by the Congressional Progressive Caucus to reduce long-run budget deficits.
The short-version of their criticism is that they examined the estimate of the growth-enhancing effect of public investment that we provided and then compared it to what would fall out of a back-of-the-envelope calculation that assumed the United States economy was well-characterized by a textbook production function. Doing this, they then decided that the implied rate of return identified in the estimates that we used was simply too high to be feasible.
Below, we’ll walk through the reasons why their judgments should not be taken as definitive on this more-technical question. First, though, we’ll say a couple of words on the odd symmetry that MA tries to paint between EPI’s analysis of the People’s Budget and the Heritage Foundation’s estimate of the 2012 Republican budget resolution (aka the Ryan Plan).
False symmetry
The (now infamous) claims made by Heritage about higher gross domestic product and lower unemployment that would result in the 10 years following the adoption of the Republican budget plan stemmed from hugely unrealistic assumptions about: (1) the response of interest rates to lower federal budget deficits, and (2) the response of private-sector investment to these lower interest rates. A key issue to note in dismissing their projections is that there is very little scope for interest rates to fall in the next decade to the degree needed to spur such large investment responses – long-term rates sit at historically low levels today.[1]
Furthermore, the Heritage Foundation’s claims about the growth-boost stemming from increased investments in the Ryan plan are even more far-fetched given the fact that the overwhelming bulk of private-sector response to lower interest rates in their model is centered on residential housing – not a sector generally thought to lead to large productivity boosts.
As a simple matter of transparency and research practice, they do not report any source for the parameters underlying their claims. Conversely, our assessment of the People’s Budget (and its public investment provisions) is based on our experience in drawing up a budget blueprint of our own. In this blueprint, the parameters concerning the possible growth-benefits of ramping-up public investment are explained in detail in an appendix and are transparently drawn from the peer-reviewed literature.
Most crucially, EPI does not use the estimates of improved growth stemming from public investments in our estimate of the People’s Budget effect on fiscal balance – for this we simply use static CBO economic projections.
Equally important, there is no symmetry between the CPC People’s Budget and the Ryan budget in terms of the substantive standards used for developing the budgets. In stark contrast to the Ryan/Heritage analysis, the revenue sources in the People’s Budget are identified, as are specific spending cuts (and increases). There are, in short, no magic asterisks. As has been well established, the Ryan budget assumes its revenue levels and the ability to substantially cut tax rates (paid for by unspecified tax expenditure reductions) and does not establish how they will be achieved. Therefore, all of the projected decreases in deficits presented in the Ryan budget are built on a significant number of these “magic asterisks.”
Macro Advisors seems frankly too eager to adopt a “pox on all houses” approach to macro-analysis of the budget debate, but this stance is not warranted, especially when their criticism of the People’s Budget and our analysis of it centers overwhelmingly on the implicit rate of return to public investment being “too high” in a very loose sense. Maybe they are right about this technical point (but probably not, see below), but this does not affect the projections of fiscal balance at all, and every step of the march toward a smaller deficit is documented in the People’s Budget but not in the Ryan plan. Furthermore, even if the quibble they have over the benefits of public investment is correct (which they themselves note is only about the size, not the direction, of its effect), then it is less a quibble with us than with what is a substantial peer-reviewed research literature on the topic. The same is assuredly not true with their (and our) disagreements with the Ryan plan and the Heritage Foundation’s analysis of it.
What bang-for-buck should we expect from increased public investment?
Now, getting to some weedier criticisms of the payoff in growth we argue could be expected from a substantial increase in public investment effort.
First, we should note again that the estimate of the growth-payoff from public investment is not actually our estimate per se – we chose one of the latest and most methodologically advanced estimates from the peer-reviewed literature (Heintz 2010) to use as our central point estimate. In fact, we were trying to be relatively uncontroversial in parameters we chose throughout the analysis of the People’s Budget – we engaged in no ‘add-factors’ or anything of the like. If the Heintz estimate was a lonely outlier in this literature, then that would perhaps raise questions, but it’s really not.
The MA response to the bang-for-buck of public investments, however, mirrors a common pattern in the empirical literature on public investment; researchers routinely find very large estimates of public investment’s effect on growth and subsequently endeavor to whittle them down to more “reasonable” sizes. Take, for example, the work of Kamps (2006), a careful study that finds large (roughly in the range of Heintz’s results) effects of public investment on growth in a time-series regression estimate for the United States. Kamps then cautions against taking this time-series estimate at face-value (for mostly sensible reasons aside from its simple magnitude), but then finds (but does not remark upon) an identical effect for the United States when examining a panel regression with 21 other OECD countries.
Second, the heart of the argument in the MA blog post is that the large implied rate of return from these regression-based findings sits uneasily with back-of-the-envelope calculations derived from a Cobb-Douglas production function; essentially saying that the simple magnitude of these rates of return is prima facie evidence that they are unreasonable. There are two responses to this: (1) these rates of return are actually far less exceptional than MA makes them out to be; and (2) it is very unclear that when one sees a possible inconsistency between (a) regression-based empirical findings and (b) the implications of assuming the United States is a Cobb-Douglas economy with constant returns to scale that it is the former which must be scrapped.
On the first point, one should be quite careful in seeing a high rate of return and assuming that its simple size rules it out bounds. Implied rates of return on public investments of ~= 40% embedded in these regression-based co
efficients are not obviously inconsistent with other findings in growth economics. Isaakson (2009) finds higher rates of return to public capital investments than this in a study of developing countries, and rates of return about half this level for advanced OECD countries as a group. But he then notes “incidentally, the estimated rates of return for the United States are approximately at par with those obtained here for developing countries.” In the private sector, Oliner and Sichel (2002) found over the 1995-2001 period that software investments yielded a rate of return of over 40% and that accumulated software investments of less than $700 billion in real dollars over that time period increased productivity by 0.35% per year – roughly equivalent bang-for-investment buck to the regression-based estimates of returns to public investments. Lastly, DeLong and Summers (1992) famously found that social returns to all equipment investment looked to be greater than 30% for a large group of countries – and that this estimate applies to rich countries like the United States as well. In short, there is nothing about a “high” rate of return that by itself invalidates the finding.
Additionally, it’s far from clear that a Cobb-Douglas production function with constant returns to scale and with no spillover effects stemming from public capital investment onto multi-factor productivity is a consistency check good enough to invalidate careful regression-based estimates. Generally, if the data doesn’t fit the aggregate production function, one should think hard about whether or not you have got the right production function to describe the actual economy.
The evidence for the United States’ production function being “Cobb-Douglas” is far from obvious, and even specifying an income or output share of public capital services is a speculative exercise. Lastly, DeLong and Summers (1992) as well as Gu and MacDonald (2008) and Perreira and Roca-Sagales (2002) have all shown that public capital investments may also improve total factor productivity and not just improve output growth through simple capital deepening.
Third, in regards to MA’s issue that large defense cuts in the People’s budget could drag on long-run productivity growth, it should be noted that there is a much thinner literature to suggest that this is worrisome. As many studies find negative effects on long-run growth stemming from defense spending as positive, and no study puts any possible benefits from defense spending anywhere in the same neighborhood as that from increases in the public capital stock.
The quibble that the list of priorities that new public investments could help meet includes things besides plain-vanilla physical infrastructure also seems less worrisome to us. Intelligent investments in many facets of education can generate clearly “super-normal” returns, for example (see Heckman’s and Robert Lynch’s work on investments in pre-K education, for example). Investments that help reduce greenhouse gas emissions also clearly pass any serious cost/benefit analysis, even if only their insurance value in reducing the probability of climate catastrophes are accounted for. And yes, if some of these investments, on their way through the political process, get turned into pure consumption, then their growth effects will be lessened.
But, on this last point that we are allegedly calling all new domestic spending “investment,” when it’s actually not, it should be noted that there’s no reason for MA to assume that the People’s Budget is smuggling in lots of pure consumption spending in the $1.7 trillion increase. The reason for this is that the $1.7 trillion increase is over and above spending levels that already assume current appropriations rise with inflation. This is already, relative to all other budget plans out there (except for EPI’s own fiscal blueprint), extraordinarily protective of social consumption spending and transfers. Because baseline consumption spending is already so well-insulated in the People’s Budget baseline, if its architects assert that the extra $1.7 billion in spending is targeted at actual investments, then there’s little reason to doubt it. So, the budget is not calling all domestic discretionary spending ‘investment’ and calculating a return on what is actually consumption not investment.
To sum up, while the precise magnitude of the growth-enhancing effects of public capital stock increases can be debated, it needs to be reiterated that the overwhelming bulk of the literature finds that these effects are positive. Our disagreement with MA is solely about the magnitude of an economic parameter and the disagreement does not affect the accuracy (or solid basis) for the projected deficits in the People’s Budget. Given that there is essentially no disagreement that public capital investments do enhance growth, the larger point remains that responsible policymaking around the federal budget is not just about bringing revenues and spending closer together but about making sure that the budget is used to enhance future living standards.
Is the People’s Budget perfect? Of course not. The most obvious way it’s not is that it targets too-aggressive a fiscal contraction (as MA rightly notes). To us, the jobs crisis should be priority number one until it is solved, and one key impediment to solving it is a premature rush to austerity. We wish the People’s Budget had reflected this view more completely, but given that political currency in the budget debate has degenerated to how fast you can target balance, it’s understandable why their deficit reduction is so aggressive. Within the universe of plans reaching near-term balance, however, the People’s Budget is a stand-out for its transparency and honesty as well as keeping an eye on the bigger picture (long-run living standards growth and economic security for typical American families) as well as just the fiscal bottom-line.
[1] Heritage also had bizarre relationships implicit between GDP, employment, and unemployment changes. For example, in 2012 they had GDP rising by $60 billion, yet employment rising by 1.1 million and unemployment falling by 2.1 percentage points. None of these three values go-together in any recognizable way based on past historical data.