Innovation and Effectiveness of Energy Policies: A Cross-National Analysis of Electricity Saving Measures in Private Households

1
Innovation and Effectiveness of Energy Policies
A Cross-National Analysis of Electricity Saving
Measures in Private Households

• Volker Schneider  Nadja Schorowsky
• University of Constance
• Transepose Interim Conference, Münster 5.11.2009

2
Contents

• Problem Strategy
• Aim of analysis
• Strategy of analysis
• Data
• Selection of countries and variables
• Data collection
• Analysis
• Basic descriptive statistics
• Cross-section analysis
• Panel regression
• Fuzzy-Set QCA
• Conclusion

3
Research Problem Strategy

• Aim
• Evaluate the contextual and political
  determinants of electricity consumption of
  private households in a large set of relevant
  countries
• Estimate the differential impacts various policy
  measures and instruments have on electricity
  saving
• Identify innovative and effective policy
  instruments, and countries with successful policy
  profiles
• Strategy Design
• Macro-quantitative Large N perspective (e.g.
  Brand Castles)
• Triangulation of different methods (Jick)
• Cross-section OLS invented in biology, diffusion
  to all
• Panel regression rooted in econometrics
• QCA rooted in comparative political science
  (Ragin)

Macro
Micro
Data
Analysis
Conclusion
Research Problem
4
Country selection based on relevance,comparabilit
y, and diversity

• 30 countries more than 80 of residential
  world electricity consumption

5
Residential electricity consumption per capita in
2006 in KWh
Abbreviations Internet Domain Names
Source UNdata own calculations
6
Dependent Variable Change in residential
electricity consumption 1995 -2006 in Percent
-7 bis 19
-20 bis 197
7
Countries and time frame of impact analysis
Source Merk (2009)
8
Policy measures and instruments
Newly Industrialized Countries without
Regional Measures, Frames and Others
9
Analysis I Cross-Section Correlations (1995-06
Data)
N30  ELCO95 ELCOG GDP95 GDPG CHDD TAX PRICE TVMW INETMW PCMW FONMW UNEMP 9506 POL-NUMB POLDIV NPXA NPX-SUM
ELCO95 1.000                              
ELCOG -0.617 1.000                            
GDP95 0.549 -0.499 1.000                          
GDPG -0.062 0.456 -0.474 1.000                        
CHDD 0.348 -0.166 0.111 0.446 1.000                      
TAX 0.155 -0.280 0.233 0.036 0.296 1.000                    
PRICE -0.185 -0.137 -0.417 0.287 0.085 0.514 1.000                  
TVMW 0.257 0.056 0.432 -0.042 -0.267 -0.066 -0.226 1.000                
INETMW 0.727 -0.613 0.801 -0.181 0.384 0.432 -0.097 0.294 1.000              
PCMW 0.707 -0.563 0.887 -0.307 0.169 0.284 -0.207 0.504 0.845 1.000            
FONMW 0.575 -0.503 0.883 -0.192 0.318 0.320 -0.196 0.267 0.831 0.794 1.000          
UNEMP-9506 -0.119 0.396 -0.287 0.422 0.172 -0.171 -0.045 0.007 -0.208 -0.408 -0.085 1.000        
POLNUMB 0.315 -0.228 0.534 -0.291 0.191 0.246 -0.204 0.225 0.518 0.421 0.457 -0.034 1.000      
POLDIV 0.150 -0.326 0.533 -0.211 0.233 0.056 -0.267 0.212 0.488 0.455 0.421 -0.429 0.231 1.000    
NPXA -0.010 0.123 0.062 0.067 0.195 -0.244 -0.429 -0.027 -0.236 -0.048 0.012 0.099 0.310 0.016 1.000  
NPXSUM 0.288 -0.180 0.469 -0.267 0.233 0.133 -0.291 0.295 0.357 0.371 0.342 -0.081 0.836 0.361 0.511 1.000
10
Analysis I Cross-section OLS regression
analysis(1995-2006 Data)

• Effects of indiv. policy instruments

• Effects of total policies

Dependent Variable ELCOG
Effect Std. Coefficient p-value
CONSTANT 0.000 0.09
ELCOSTOTSHARE -0.415 0.00
GDPG 1.029 0.00
CHDD -0.662 0.00
NPXA 0.261 0.06
NPXCL 0.051 0.63
NPXCS 0.022 0.86
NPXE 0.073 0.51
NPXI 0.217 0.096
NPXVL -0.028 0.828
NPXVS 0.181 0.151
N 30
Multiple R2 0.878
Adjusted Mult. R2 0.814
Dependent Variable ELCOG
Effect Std. Coefficient p-value

CONSTANT 0.000 0.02
ELCOSTOTSHARE -0.400 0.00
GDPG 0.882 0.00
CHDD -0.496 0.00
PRICE -0.057 0.51
TVMW -0.083 0.36
NPXSUM 0.453 0.00
N 30
Multiple R2 0.860
Adjusted Mult. R2 0.823
11
Analysis II Panel regression analysis
(Merk)(1995-2007)

• Christine Merk is testing several models with
  lagged dependent variable, control variables
  (GDP, climate, etc.), policy variables,
  different country groups (all countries AC,
  high income countries HIC, third model incl.
  prices)
• Selected regression results for policy variables
  mostly dummies ? with standardized regression
  coefficients
• AC Model 3 incl. prices
• Agency? -.034 -.256
• Information material? .080 .088
• Campaign? -.014 -.013
• Energy counselling? -.081
• Smart metering? -.033 -.108
• Price -.015
• Subsidies scheme ? .071 .051
• Tax incentives ? .029 .072
• Low interest rates ? -.131 -.027

Legend N AC 330 N HIC 197 R2adj. .999 plt0.05
plt0.01 plt0.001
12
Energy efficiency agencies
Source Merk (2009)
13
Analysis III Fuzzy-Set QCA 1995-2006-Data
(Mayer)

• QCA is looking for constellations of causal
  conditions, i.e. presence and absence of
  determining factors or policy measures
• Crisp-set QCA ? Boolean truth tables ? needs
  binary values
• Fuzzy-set QCA ? analysis of ordinal and metric
  data
• Aim is, to find causal constellations (? model)
  that explain an outcome which is covered by a
  number of cases
• Ines Mayer is testing a number of models such as
• elco f(climate, cross, oek, info, coop, legal)
  and finally finds a solution in
• which contextual conditions have a large
  explanatory power
• raw coverage unique coverage consistency
• climate ? gdp 0.53 0.09 0.87
• urban ? gpd 0.54 0.10 0.83
• solution coverage 0.63 solution
  consistency 0.84

14
Fuzzy set Policy model

• Model elco f(cross, oec, info, coop, leg)
• Assumptions cross (present) oec (present) info
  (present) coop (present) leg (present)
• Raw coverage unique coverage consistency
• coop ? ? leg 0.30 0.05 0.87
• cross ? info 0.52 0.14 0.89
• cross ? ? oec ? leg 0.41 0.10 0.82
• solution coverage 0.68 solution
  consistency 0.80
• Conclusion Cross-Policy measures (e.g. energy
  efficiency agency or fund) explain many cases

15
Policy-Effectiveness based on QCA classification
Low Growth
High Growth
Adapted from Mayer (2009)
16
Conclusions and implications

• Several limitations affect interpretation and
  generalizability, however, some results are
  promising
• Energy agencies are important to increase
  awareness
• Metering, energy councelling and information
  campaigns have negative impact on electricity
  consumption
• Economic incentives mixed Low interest rates
  with negative impact
• QCA results underline findings
• Cross-section instruments combined with
  informative measures always have an impact
• Economical and regulatory instruments seem to
  have no specific effect
• Danmark is the most successful country based on a
  broad efficiency policy profile

Data
Analysis
Conclusion
Research Problem
17
References

• Brand D, Saisana M, Rynn L, Pennoni F, Lowenfels
  A. 2007. Comparative analysis of alcohol control
  policies in 30 countries. PLoS Medicine 4752.
• Castles F. 2003. The world turned upside down
  below replacement fertility, changing preferences
  and family-friendly public policy in 21 OECD
  countries. Journal of European social policy
  13209.
• Jick T. 1979. Mixing qualitative and quantitative
  methods Triangulation in action. Administrative
  science quarterly602-11.
• Mayer I. 2009. Energieeffizienz in privaten
  Haushalten im interntionalen Vergleich. Eine
  Policy-Wirkungsanalyse mit QCA. Master Thesis.
  University of Constance.
• Merk C. 2009. The impact of energy efficiency
  policies on residential electricity consumption.
  Master Thesis. University of Constance
• Ragin C. 2000. Fuzzy-set social science. Chicago
  University of Chicago Press.
• Ragin C. 1989. The comparative method Moving
  beyond qualitative and quantitative strategies.
  Berkeley Univ of California Press.

18
Abbreviations

• TVMW Mean level of TV penetration 1995-2006
• NPXA policy index agencies
• NPXCL policy index compulsory labeling
• NPXCS policy index compulsory standard
• NPXE policy index economic incentives
• NPXI policy index information measures
• NPXVL policy index voluntary labeling
• NPXVS policy index voluntary standards
• NPXSUM Sum of policy index
• PCMW Mean level of PC penetration
• POLNUMB Number of policies
• POLDIV Diversity of policies
• UNEMP-9506 Unemployment change

• ELCOSTOTSHARE residential consumption as a share
  of total electricity consumption
• CHDD Mean yearly cooling and heating degree days
  1995-2006
• ELCO95 residential electricity consumption in
  1995
• ELCOG Mean growth rate of residential electricity
  consumption in 1995
• FONMW Mean level of Phone penetration 1995-2006
• GDP95 GDP in 1995
• GDPG Mean growth rate of GDP 1995-2006
• INETMW Mean level of internet