Looking at general scenarios is useful to give some understanding of the role of energy efficiency for future energy requirements. But to understand the process of efficiency improvements and how they come about, we must recognize and understand why energy efficiency change does not proceed in a linear fashion. Analysis by Nakicenovic (1997) of historical technological change illustrates how economic activity undergoes paradigm shifts based on successive long waves of key technologies. The paper illustrates how the process of economic growth and technological change is not smooth and continuous and that the timing of the transition from one dominant cluster to another is consistent with the pattern of Kondratieff long waves. Energy efficiency has been shown to respond in a Hicksian relationship to the cost of energy (Newell et al., 1999): when energy prices have risen, then energy efficiency has also risen. Lastly, there is an inherent difficulty in assigning energy use to technologies and activities that have not yet been developed or even conceived. For example, what was the energy efficiency of personal computing technology 20 years ago, never mind 100 years ago?
A key distinction between the high and low estimates of future energy intensity is the projected cost of energy. Simply put, when energy is cheap, there is little incentive to save it. But price is clearly not the only factor. Tax structure, R&D investments, efficiency standards, and innovative efforts to commercialize new technologies are also important elements. Public policies that correctly price energy by internalizing externalities and help to overcome market failures can go a long way towards leveling the playing field for efficiency improvements to compete with supply options. However, the history of energy policy has often seen subsidies of less-than-optimal energy supply technologies and fuels supported for a variety of largely political motivations. Also important in energy efficiency is the market and regulatory structure of energy supply and information flows concerning energy options. Lastly and crucially, social processes provide inertia against changing energy-intensive practices, and social dynamics can result in unanticipated increased energy usage. The classic US (and other parts of the developed world) example of this is the dependence on the automobile, and the related exodus of people from inner cities to suburbia and onto exurbia.
Empirical analysis of historical energy use has tended to focus either on aggregated energy intensity in a top-down approach, or technology-specific energy efficiency developments in bottom-up studies. As described in the previous section, there is considerable difficulty in comparing and translating results on intensity to efficiency and vice versa. Generally, bottom-up analyses have shown larger improvements in the effective use of energy. This is at least in part due to the fact that technical efficiency gains have been devoted to enhancing energy service attributes rather than reducing energy inputs needed to secure a given level of service. For example, technology gains in automobile efficiency have gone to providing more vehicle power and other amenities rather than to getting more miles to the gallon.
We have stressed the caution needed when considering long-term trends. Energy efficiencies for specific technologies or activities can be defined. But aggregating these measures to an economy's energy intensity is further
Energy Consumption per Dollar of Gross Domestic Product
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995
Figure 4.2 Estimated US energy intensity 1950-1996 (Source: EIA, 1997).
complicated as the shifting mix of economic activities in any country changes over time.
With these difficulties in mind, we begin by looking at estimates of past energy intensities in the US. The World Energy Council reports the long-term rate of energy intensity improvement in the US, averaged over the past 200 years, was about 1% per year (WEC/IIASA, 1995). However, analysis on decadal time scales (Dowlatabadi and Oravetz, 1997) has shown that energy intensity responds to changes in energy prices beyond the price elasticity of demand. Thus for the 20 years from 1954 to 1974, energy intensity rose by close to 2% per annum. This can be attributed to technological drift as energy efficiency was not incorporated into the design of goods and services. Following the 1970s oil shocks, the energy intensity of the US economy declined in response to price signals; from 1979-1986, energy intensity fell an average of 3.1% per year. However, as energy prices tumbled post 1986, the decline in energy intensity steadily slowed as the US failed to carry on the energy efficiency experiment long enough to see continued improvement (Rosenfeld, 1999). Figure 4.2 shows another estimate of US energy intensity trends.
In other industrialized nations, a similar story is seen. Following the two oil crises in the 1970s, energy intensity over the years 1973-1993 declined by 1.4%/yr in the IEA countries (IEA, 1995). Following the collapse of energy prices post 1986, the rate of decline slowed, reaching 0.6%/yr on average for the IEA countries over the period 1989-1993. Thirteen IEA countries actually report increasing E/GDP trends (WEC/IIASA, 1995).
So what drives trends in long term energy intensity? The authors of
WEC/IIASA (1995) conclude that "energy intensity improvement rates are related to per capita GDP growth rates. The faster an economy grows in per capita terms, the faster is productivity growth, the rate of capital turnover and the introduction of new technologies, and the faster energy intensities improve." However, an alternative result of economic growth can be greater investment in energy intensive industries, and subsequent increases in energy intensity as long as saturation for energy intensive goods has not been reached. Conversely, in cases of negative per capita GDP change, such as the recent experiences in the former Soviet Union, energy intensities decline, even though overall energy use may also be declining. This latter effect dominates for Kazakhstan, as is shown in Figure 4.4.
Dowlatabadi and Oravetz (1997) also find that intensity trends have varied widely through time and responded strongly to prevailing economic conditions. They postulate that scrapping of old capital plays a key role. This can often come about as a precursor to the growth phase of the economic cycle and is associated with deep recessions. The deep recession leads to the death of activities and behavior that is economically of marginal value (this is despite public policy attempts to save jobs, industries, or communities for political motivations), and is often characterized by energy inefficient capital in sunset industries. This structural change in economies opens the path for new ways of meeting demands and new behaviors demanding new energy services.
Given that declines in energy intensity are non-linear and respond to price changes and structural shifts in the economy, what evidence is there for individual energy efficiency improvements to take advantage of new opportunities? Ausubel and Marchetti (1996) discuss ongoing historical trends in energy efficiency improvements from an engineering perspective. They point to the 300-year quest to develop more efficient engines, from 1%-efficient steam engines in 1700 to today's best gas turbines which approach 50% of their theoretical efficiency limit (Figure 4.3). Fuel cells, which Ausubel (1998) says may power our cars in 20 to 30 years, will increase that efficiency to about 70%, as fuel cells do not incur the inevitable efficiency limits of combustion systems imposed by the laws of thermodynamics. Similarly, Ausubel and Marchetti (1996) point to the dramatically increasing efficiency of lighting technology over the past 150 years (Figure 4.3, analyzed as a sigmoid (logistic) growth process, shown in a linear transform that normalizes the ceiling of each process to 100%).
Of course, technological change can also lead to worsening energy efficiencies. For instance, in the post-WWII era, rising incomes and falling electricity prices led to the marketing of refrigerators with less insulation and larger interior volumes (von Weizacker et al., 1995).
Figure 4.3 Improvement in the energy efficiency of motors and lamps (Source:
Ausubel and Marchetti, 1996).
1700 1750 1800 1850 1900 1950 2000
Figure 4.3 Improvement in the energy efficiency of motors and lamps (Source:
Ausubel and Marchetti, 1996).
On the whole, however, empirical evidence exists for considerable energy efficiency improvements in specific technologies. In some technical areas the improvement has been remarkable, for example in energy efficient lighting, fuels cells and photovoltaics. As impressive, and potentially even more important, is the corresponding decrease in costs of these energy efficient technologies. However, it should be noted that there is an inevitable time lag for adoption of efficiency improvements (in addition to reluctance of potential adopters as discussed later) due to the delay of capital stock turnover. As Rosenfeld (1999) explains, capital stock (for a proxy life of 15 years) begins to turn over the first year, but the improvement in energy efficiency is then just the fraction of new units multiplied by the fraction of these that were higher efficiency units, e.g., 5% of 5%. These fractions grow each year until the stock is completely replaced.
To continue our discussion of how improvements (or lack thereof) in energy efficiency from individual technologies are intricately tied to structural changes in the economy, we must make the distinction between models in which tech nological learning is exogenous, or assumed in the model, or the far more realistic cases in which it is endogenous, or affected by other parameters in the model.
In traditional energy-economic models, energy intensity is exogenously specified, that is, it is dealt with as an input which is not responsive to anything within the system. In macroeconomic models (Manne and Richels, 1992; Walley and Wigle, 1991), this process is captured by the autonomous energy efficiency improvement (AEEI). The AEEI is assumed to capture all non-price induced changes in energy intensity, including structural change in the economy and sector-specific technological change. Thus, it is neither a measure of technical efficiency improvements per se, nor is it autonomous since policy-driven adoption of more energy efficient technologies is included. AEEI is usually specified in a range between 0.5% and 1.0% per annum, although some modelers use even lower values. However, the empirical evidence and theoretical understanding of technological change does not support such a relentless linear efficiency improvement (or decline in energy intensity).
A more realistic categorization for energy intensity makes it endogenous, or responsive to the interacting system factors such as policies, prices, research and development, and path dependence (Goulder and Schneider, 1998; Dowlatabadi, 1998). Generally, results from these models are consistent with high energy prices leading to a demand pull model of technological change improving the efficiency of energy use. In times of low energy costs, other characteristics including performance and ease of use may be valued more than efficiency.
Was this article helpful?