Building Codes That Work
If I got a dollar each time someone says that California’s energy efficiency codes have led to significant decreases in electricity consumption, I could buy a Tesla to help reverse that trend. In the halls of power, climate regulators discuss this source of energy savings potential with a level of excitement rivaling that surrounding the appearance of Harrison Ford as Han Solo in the new Star Wars trailer. New building codes are a significant part of projected emissions reduction goals in the US, Europe, Japan and elsewhere. The question of course is, whether building codes actually cause such decreases in energy consumption.
Figuring out the realized magnitude of energy savings from building codes is tricky. You cannot just compare the energy consumption of buildings built today (post building code) to those built prior to the imposition of building codes for at least three reasons:
- Today’s homes are much bigger and we are increasingly building new homes in hotter parts of the state/country.
- People who use a lot of energy consuming services (e.g., cooling) might self select into more efficient newer homes.
- The imposition of a building code is not random, but a policy choice. Areas with extreme seasons and a greener populace are more likely to adopt such regulations.
There are a number of papers that have tried to overcome some of these barriers. Yours truly tried to overcome the problem of policy endogeneity (problem 3 above) in a cross-state and -time comparison and found savings of about 2-5%. Kotchen and Jacobsen in a nice experiment compare new buildings pre and post a building code in Gainesville, Florida and find savings of a similar magnitude. These savings are significantly smaller that ex ante engineering estimates, yet still economically and statistically significant.
A recent NBER working paper by Arik Levinson, who recently worked for the White House Council of Economic Advisors, argues that there is no evidence that California’s building codes have led to a reduction in electricity consumption after you address the three issues above. This paper struck a nerve with my friends in Sacramento and was featured on the Freakonomics podcast. So what does it do?
Using data from two rounds of California’s Residential Appliance Saturation Survey for about 16,000 homes matched to detailed electricity billing data, he estimates regressions, which account for detailed characteristics of the homes and occupants and the climate zones they are in. The key variables of interest are indicators of year built for each housing unit. He finds no statistically significant evidence that buildings of younger vintages use less electricity than older buildings – with the exception of the most recently constructed buildings.
Levinson then questions this finding for the most recent years. What if buildings become leaky after just a few years? Or maybe new homeowners have no money to spend on electricity and conserve energy right after purchasing a new home. As time goes by and budgets become less tight, they just might turn on the AC more frequently. Figure 3 in the paper makes exactly this point.
What you see here is electricity consumption against building age by construction decade. The fact that the leftmost line segments slope upward most steeply suggests that newly built houses within a construction cohort do consume less electricity. Levinson argues that this is in fact evidence in favor of the point that buildings deteriorate quickly after being built and/or residents turn up the heat/AC once they have more cash.
The paper also shows convincing evidence that buildings built under different building code regimes do not have statistically different temperature response profiles. He digs into national survey data and shows further evidence in support of his findings based on California data.
If you stop reading and thinking here, you might walk away with the idea that building codes are useless and we should spend our money on more worthwhile causes like desalinization (don’t get me started on that bright idea). Don’t walk. One more paragraph. You can do it.
The paper recognizes up front that owners of older homes might spend money on insulation, new windows and better sealing to make their homes more efficient. This would of course make the older pre-code homes more like post-code homes and increase the likelihood of a no effect finding. Does this happen? A three thousand dollar rebate check on its way in the mail to me from Sacramento for my newly sealed 1947 built home is evidence that this happens. Even my politically conservative neighbors have been spotted with the insulation truck outside their 1948 home.
Second, we will never observe the true counterfactual home that would have been built instead of the building code compliant home that was eventually built. Even the most careful econometrician does not observe all relevant characteristics that change over time.
Third, building codes provide many benefits that are not measured in kilowatt-hours, but in comfort. Visit your strange friends living in a house with single pane glass and sit near the window on a cool night.
Finally, much of the benefit from building codes comes from lower natural gas consumption for heating. The paper does not study this dimension in great depth.
Arik, who is an incredibly careful and thoughtful economist, is careful in discussing all of this in his paper, but he still comes to the conclusion that building codes do not result in savings that should be counted as additional under new climate and energy regulation. The main argument there relates to the fact that if people in older homes voluntarily improve the efficiency in their homes, then building codes simply build this into the up front cost of a new home. This makes the new building code essentially a choice that people would make in the absence of the policy eventually and is hence not additional. There is some economic truth to this argument.
In order to settle this, I am afraid, we would need to run one really expensive RCT, where some identical homes are built according to building codes and others are not. We would then have to have random people assigned across these homes and live their very real lives in these homes. If you are a developer, give me a call. I am standing by having a hot cup of tea in my now comfortable, no-longer-leaky California home.
Maximilian Auffhammer View All
Maximilian Auffhammer is the George Pardee Professor of International Sustainable Development at the University of California Berkeley. His fields of expertise are environmental and energy economics, with a specific focus on the impacts and regulation of climate change and air pollution.
Seven of eight years ago we replaced the windows in our 1970s vintage home on the Peninsula with dual pane windows. Under the most optimistic of circumstances (assuming 2008 gas prices) the payback was 20 years. The principal benefit was a reduction in ambient noise.
Five years ago we built a home in the mountains that is half again as large as the one we left behind in a climate that is three times as cold as measured by heating degree days. Our natural gas bills are one third lower in the mountains than they were in the Bay Area. Part of the difference is better building codes and part is the use of much more efficient in-floor radiant heating that also allows better temperature management of unoccupied space. I’m not sure how heating efficiency with the old-fashioned radiators compares with in-floor heating but it has to be far better than forced air, which in our experience was never very comfortable at any reasonable temperature setpoint.
Max, help me understand Levinson’s interpretation of Figure 3. California’s building codes are mostly aimed at reducing space heating, which in Figure 3’s homes are 100% fueled by natural gas. Should we expect to see any impact on electricity use? Well, yes, the furnace fan will operate less (but that’s a tiny effect). What about less AC electricity? Well, yes, but AC electricity is responsible for 4% of California’s residential electricity use, so it would take an increase of several hundred percent to cause that observed change. In any event, “increased leakiness” is a laughable explanation. Meanwhile, residential natural gas use–the stuff actually targeted by building codes– has fallen by roughly half over the period discussed. Am I missing something?
In fact, I suspect Figure 3 is mostly highlighting the increase in the “other” uses of electricity– a trend that is totally unrelated to building codes.
gas data: http://energyalmanac.ca.gov/naturalgas/residential_natural_gas_consumption.html
Without spending a lot of time on this, isn’t he taking into account home size at all? Newer homes have steadily gotten larger. The trend is there well before the codes first went into effect (~1980) so how can he argue that codes would have had ANY effect before then. There is just too much muddiness in this analysis.
The graph shows a secular trend of a 30% reduction of energy use measured at the date of complete, the midpoint or the end point. Because each cohort started lower, they ended lower. That is independent of their behavior. Moreover, the graph does not show the increase in house size, which was 40% over this period, so the energy consumption per square foot declined, on a secular basis by over 60%.
Two thoughts: It’s not clear if the graphic is total usage per home or usage per square foot. The latter measure would be useful to disaggregate the effects of building codes against rising income which leads to bigger houses.
Which leads me to the second thought–what’s the impact of plug load such as computers and TVs on this measure? Appliances would be the other aspect to pull out since building codes don’t affect those.
Building codes should primarily affect heating and cooling. In Levinson’s paper, Figure 4 shows that over 90% of homes of all vintages do not have electric heat. According to the Figure you show, the analysis is about houses without electric space or water heating. So the conclusion that building codes do not decrease electricity consumption is not surprising. Wouldn’t larger homes, with increasing saturation of air conditioning and far more electronic end uses over time explain the increase in electricity consumption over time, regardless of what the building codes are? Consumption of natural gas, the dominant fuel in California for space and water heating, is shown to decline over time.
These are interesting data but, as you note in a rather understated way, they are moderately to potentially highly misleading in terms of trying to understand what the effects are of building codes. According to the EIA summary compilation of CA residential energy use (found here: http://www.eia.gov/consumption/residential/reports/2009/state_briefs/pdf/ca.pdf) about 44% of 2009 end use energy consumption in CA is due to “appliances, electronics and lighting” (these are apart from end use energy for air conditioning though without looking more carefully at the data, I’m not sure how furnace fan electricity consumption is apportioned). Presumably almost all of this 44% is electricity. Its hard for me to see how BUILDING CODES have a big effect on electricity consumption for non-heating/cooling loads – maybe the requirement to use fluorescent lighting in kitchens is an example of a building code requirement? But clearly for homes heated with natural gas, as ~60% of CA homes are, the effects of requiring (more) insulation, better windows, set-back thermostats, etc. are going to more apparent in the natural gas consumption data.
For non air-conditioned homes in CA, annual natural gas consumption is several times higher than electricity (I’m not sure how they compare in an air-conditioned home, but the EIA data suggest that air conditioning only accounts for 4% of end use energy consumption). So rather than trying to tease out a secondary effect (furnace fan electricity consumption?) from the electricity usage data, wouldn’t it be better to look at NG data directly (as you note: “Finally, much of the benefit from building codes comes from lower natural gas consumption for heating. The paper does not study this dimension in great depth.”).
While it is indeed hard to run the ‘case-control’ study to see what the comparison would be between having building codes for energy and not having codes (especially longitudinally), I would think there might be comparative data from other states where building code requirements were implemented later than in CA, selecting for climate zones and other demographics…
One point of agreement – the benefit of retrofitting old single pane windows with double glazed windows has, as you suggest, quite noticeable changes in comfort – especially for noise reduction.
Finally, a couple of nit-picks on your “figure 3” – the units on the y axis are MBtu/year/dwelling unit (not obvious until you convert to kWh) and I’m not sure I understand data plotted before t = 0 ??
Agreed, a good post. Like many regulations that rely on evaluation against a counterfactual, it seems that EE program success will consistently be fodder for talented econometricians. Funny how an appropriate price for social damages (for all production and consumption costs) could obviate the need for most of these debates, save of course potential market failures seen with rental properties.
One specific question though–Politically conservative neighbors? In Berkeley? Clearly you must live out in Walnut Creek, , , , ,
Maximilian — Areas without building codes exist within the US, and I believe if you looked at those areas you would probably find two things: 1) poor people live there and don’t cannot afford the extravagance of burning fossil fuels to pollute our planet — so they don’t use much electricity and 2) that as a whole correcting for income, the buildings use more energy than those in states with progressive codes. John Proctor, P.E.
Good blog post, Max. I would add that as consumers we regard energy efficiency savings as a bonus to be spent on something we want and now have earned by our efficiency savings. That is why even though newer home construction is larger than older homes energy intensity may actually be equal or less. As consumers we spend our energy efficiency savings on a bigger TV not some politician’s ‘do gooder’ wish list.