Rate this post

I have a new functioning paper out, coauthored with Stephan Bruns and Alessio Moneta, titled: “Macroeconomic Time-Series Proof That Power Efficiency Improvements Do Not Save Power”. It really is yet another paper from our ARC funded project: “Power Efficiency Innovation: Diffusion, Policy and the Rebound Impact”. We estimate the economy-wide impact on power use of power efficiency improvements in the U.S. We locate that the rebound is about 100%, implying that in the extended run power efficiency improvements do not save power or minimize greenhouse gas emissions.

At the micro level, we may naïvely count on a 1% improvement in power efficiency to minimize power use by 1%. But folks adjust their behavior. Efficiency improvements minimize the price of power solutions like heating, transport, or lighting. Mainly because these are now less expensive to create, folks consume much more of them, and so the percentage reduction in power use is significantly less than the improvement in efficiency. This is identified as the direct rebound impact.

People today may also redirect their spending to consume much more of complementary goods, like bigger homes in the case of residential heating improvements, and minimize their consumption of substitute goods and solutions, like bus rides or cycling, in the case of car or truck fuel economy improvements. These modifications have implications for the power utilised to create these goods and solutions. In addition, the reduction in demand for power must reduced the value of power additional boosting the rebound in power use. Ultimately, the improvement in power efficiency is an boost in productivity, which must outcome in financial development. Larger incomes imply larger demand for power. Adding these indirect rebound effects to the direct rebound impact we get the economy-wide rebound impact.

The size of the economy-wide rebound impact is vital for estimating the contribution that power efficiency improvements can make to lowering power use and greenhouse gas emissions. Our study delivers the 1st empirical basic equilibrium estimate of the economy-wide rebound impact. Earlier research use simulation models, identified as computable basic equilibrium models, or partial equilibrium econometric models that never enable the value of power to adjust. Some of the latter research also measure rebound incorrectly, for instance assuming that power intensity – power utilised per dollar of GDP – measures power efficiency. In reality, the majority of the rebound impact takes place when power intensity rebounds as folks shift to much more power intensive consumption following an power efficiency improvement. Financial development induced by the efficiency improvement is anticipated to contribute significantly less to total rebound.

We use a structural vector autoregressive model, or SVAR, that is estimated making use of search strategies created in machine finding out. We apply the SVAR to U.S. month-to-month and quarterly information. An SVAR explains modifications in the vector of variables, x, in terms of its previous values and a vector of serially and mutually uncorrelated shocks, ε:

In our fundamental model, the vector, x, includes 3 variables: main power use, GDP, and the value of power. The 1st of the shocks is a shock to power use, holding continuous shocks to GDP and the value of power and the previous values of all 3 variables. We believe this is a affordable definition of an power efficiency shock. The other two shocks are revenue and value shocks.

The matrix, B, which transmits the shocks to the dependent variables can not be estimated with out imposing some restrictions or situations on the model. Normally economists use financial theory to impose restrictions on the coefficients in B (quick-run restrictions) and the Π_i (extended-run restrictions). Alternatively, they sample a variety of models, rejecting only these that never meet qualitative “sign restrictions” on the matrix B. Alternatively, we use independent element evaluation, an strategy that is fairly new to econometrics. This imposes situations on the nature of the shocks as an alternative and estimates B with out direct restrictions. In contrast to the quick- and extended-run restrictions strategy, it does not impose a priori restrictions on the information, and as opposed to the sign restrictions strategy, it estimates a distinctive model.

Employing the estimated SVAR model we compute the impulse response functions of the dependent variables to the shocks:

The best left graph shows the impact of an power efficiency shock on power use. The grey shading is a 90% self-confidence interval, the x-axis is in months, and the y-axis in log units.

Initially, an power efficiency shock strongly reduces power use, but this impact wears off more than the following years as customers and the economy adjusts. At some point, there is no adjust in power use so that rebound is 100%.

The other graphs in the 1st column show the impact of the power efficiency shock on GDP and the value of power. The second column shows the impact of a shock to GDP, and the final column an power value shock.

The implications for policy are that encouraging power efficiency innovation is unlikely to make a contribution to lowering greenhouse gas emissions. This is one particular purpose why I am skeptical of projections that predict that power intensity will fall significantly quicker in the future than in the previous simply because of power efficiency policies.

On the other hand, if these policies raise rather than minimize the fees of creating power solutions then the direct rebound (and presumably the economy-wide rebound) will be adverse rather than optimistic. As, apart from their environmental effects, these would minimize financial welfare, it appears that there would be far better selections to minimize emissions by switching to low carbon power.

Source link