Does Washington Listen?
Data mashup time. I’ve previously posted time trends of 40 years of public opinion data on government spending on social issues, as well as 50 years of actual government spending (more or less). Now it’s time to see whether the two actually track. Does Washington actually listen to public opinion?
The graphic below sets the template for the detail plots below. The solid grey curve shows the net support for social issue spending from the General Social Survey. (In the example graph, I use the crude composite score from my previous GSS post, just for purposes of illustration. Issue by issue scores will be used below.) This curve shows [percentage of people who say too little is being spent] minus [percentage of people who say too much is being spent].
The green bars show (smoothed) year to year changes in Federal spending on various issues (in the example graph, it’s just the total budget). If spending is increasing from one year to the next, the bars will be positive above the (invisible) y-axis above; if spending is decreasing, they will be negative. Technically the bars represent changes in spending expressed as percentages of GDP, rather than absolute dollars, so more properly, they denote changes in Federal “investment priority”.
Now, a caveat. Because of the mechanics of the three years it takes to prepare, request, and appropriate Federal budgets, as well as the two-year granularity of much of the General Social Survey data, it isn’t really possible to show cause and effect here. The best we can say is whether the “coupled public opinion / spending system” is roughly tracking, or out of whack, in any given year. Even with those data limitations, you’ll see some surprisingly coherent behavior below.
One final note: to preserve design minimalism, I’ve followed Tufte rules and stripped the graphics down pretty heavily. In all graphs below, the data use the same scaling. The public opinion curve spans a dynamic range from +80% to -80%, and the year to year spending changes span a dynamic range of +20% to -20% of their then-year levels.
OK, on to the data.
I’ll start with Space Exploration, both because I’ve written about it before as well as the fact that the trends are spookily close. In the GSS, public opinion is asked on the Space Exploration Program. For federal spending, I’ve shown the Space Flight Subfunction of the budget tables. The signal is dominated by parts of NASA’s budget. The storyline here is actually, I think, fairly straightforward: space exploration has never been a net positive spending support category, but the antipathy built up during Apollo era spending decreased during the 1970s. Major successes (delivery of the Space Shuttle in 1981, and post-Challenger Return To Flight in 1988) were coincident with relative increases in space flight spending. Conversely, the early 1990’s saw the Hubble mirror failure, GHWB’s unpopular and expensive exploration program, and the very public and bruising battles around the International Space Station. Again, cause and effect are indeterminate, but it is clear that a period of falling public support overlapped with a period of falling national investment priority. What is fairly remarkable is that consistently, periods of increasing support correspond to, at the very least, periods of reduction in the downward pressure on space spending, and very occasionally, to increases. Since the overall trend (as share of GDP) is nonetheless negative, this seems to perhaps support ex-NASA-Administrator Mike Griffin’s “next fifty years” proposition, which is that space spending has been, and is likely to, remain more or less flat in absolute dollars (rather than in investment priority / share of GDP).
In this chart, public opinion is for Military, Armaments and Defense, and spending data are for the National Defense Superfunction of the budget tables. The public opinion data for this category are fascinating: they tip positive after instances of direct acts of aggression against the U.S. (1980 – Iranian hostage crisis; 2001 – 9/11 attacks). Although just as interestingly, the upward trend in support in both cases had been nearly a decade long. While there is not coherence in phase between opinion and spending, overall, the patterns are more synchronized than contradictory. What I think I see here is a consistent lag between spending changes and public opinion, probably related to the sheer size and inertia of defense program budgeting, appropriate and spending.
Public opinion here is for Improving and Protecting the Nation’s Health, and spending data are for the Health Function of the budget tables. There’s certainly no question that (up until recently) Americans have been in violent support of health care spending. Periods of comparatively modest increases in the strong “baseline” support definitely tracked with periods of increased national investment; conversely, declines in support corresponded to periods in which spending increases slowed, but rarely reversed. The 2010 downturn in public support, presumably in response to the national dialogue on the subject, is unprecedented in the GSS historical record; it will be fascinating to see which way the investment trend heads.
Here, the data are opinion data on Improving and Protecting the Environment, and spending data for the sum of the Pollution Control and Abatement Subfunction and the Conservation and Land Management Subfunction. Some of these patterns now look familiar: support for more spending increasing through the 1980s, peaking around 1990, followed by a fall in support along with (or after) periods of actual increased spending. It would be tempting to speculate a cycle in which cuts in this area generate greater support for restored funding, which builds in intensity and eventually overturns the reductions and turns them into increases … after which support drops. Now again, that would be reading too much into the data since we can’t correlate causality here, but … it’s not an impossible storyline.
The opinion data here are for Halting the Rising Crime Rate, while the spending data are for the Administration of Justice Function of the budget tables. This is an instance where the public opinion trend may be out of sync with the spending trend; while overall strong, public opinion support for spending on crime has fallen since the early 1990s, while Federal spending on the justice system has continue to grow with almost no interruptions. In fairness, it is difficult to make a direct intercomparison here, as most spending on crime is at the local, rather than Federal, level (the same will be true for Education, below).
The data here are for Improving the Nation’s Education System (public opinion) and the Elementary and Higher Education Subfunctions of the budget tables (spending). As with crime, the trends do not closely track each other. It is tempting to speculate that this is due to the local vs Federal issue; in both cases, both opinion and spending may be tracking broader policy debates around appropriate Federal roles, rather than showing a closely coupled system in which acknowledged Federal spending functions are in sync with and reflect public sentiment. In both cases what is really needed is a long term trend of the national (not just Federal) public sector spending towards these issues.
Turning from the solidly popular to the manifestly unpopular, the data here track Foreign Aid (public sentiment) and the International Development Aid Subfunction (spending). Here, the slight uptick in public support after 2000 (while still overwhelmingly net-negative) corresponds with significant increases in international development aid.
I have to say, overall these data have more hallmarks of a “better behaved system” than I was really expecting. It will probably never be possible to untangle the hairball that is policy, public opinion, and spending from a causal perspective, but it is reassuring to see even indirect evidence that the system self-corrects in practice (spending), rather than just appearance (politics and elections).