Economies of scale and utilization swiss electricity distribution industry

Economies of Scale and Utilization in the Swiss Electric Power Distribution

Industry

Author: Massimo Filippini

Presenters: Arvind K. YadavRashi Saxena

Electric Utility Industry

• Local monopolies• Cyclical demand• Require spare capacity for peak

periods– Too high?– Over-capitalization?

• Swiss electricity distribution industry– Economies of scale empirically evident

Cost Structure

• Convention: long-run cost functions• Implication: – static equilibrium– Optimal utilization of inputs

• Contention: Absence of static equilibrium w.r.t. stock of capital (quasi-fixed)

• Implication: Economies of scale based on LR cost function may be imprecise

Suboptimal capacity: Supporting arguments

• Costly adjustment to time profile of electricity demand– Longevity of transformers and distribution

lines– Long-term load forecasts and distribution

planning (inaccurate)

• Legal obligation to maintain excess capacity– Service guarantee– Exclusivity of territorial franchise

Variable Cost Function

• To model production structure• Takes account of sub optimality• Physical capital can’t be adjusted to

minimize TC during observation period

Production function …y= F(x1,x2,…, xg; k1,k2,…ke; q1,q2,…qn; T) ❶

– Y: output; x: inputs; k: quasi fixed inputs; q: operating and o/p characteristics variable;

T: vector of time shifts

Variable Cost Function

Properties• Concave and linearly

homogenous in i/p prices

• Non-decreasing in input prices

• Decreasing in quasi fixed inputs

Inputs• Labor• Purchased power• Quasi fixed input

capital

VC function of a Swiss electricity distribution utilityVC = VC (y, wp, wl, k, T, LF, FDj) ❷

- y: output (kWh); wp, wl : kWh i/p and labor prices-K: stock of capital; T: time; LF: load factor; FD: firm specific variables

Translog function

Sl = βl + µu ln (wl/wp) + ωyl ln y + πlk ln k + ᵟlLF ln LF ❸

• Tested for– Homotheticity– Cobb-Douglas technology

Data/Structure

• Swiss electric power industry• 1200 firms (public/private)– 900: municipals– 300: urban/regional

• Generation/Transmission/Distribution: small amount of power generated

• 10 main utilities vertically integrated– Generation/Transmission/Distribution:

backbone

• 74%: distribution utilities

Caveats and Procedure

• Publically owned• Data available: 60 utilities• Utilities with more than 20% of their capital

invested in generating activities (21 nos.): excluded39 distribution utilities serving cities were

analyzed• Measures of capital stock:– Capacity measure– Cost measure (data N.A.)

Results

• Four models:–Model 1: Estimates of VC function model

specified in ❸–Model 2: homotheicity assumed–Model 3: corresponding to Cobb-Douglas

technology–Model 4: ignores firm-specific effects,

biased

• Statistically significant coefficients• Cost elasticity < 1

Results

Expected• VC function should be

increasing w.r.t. output and input prices

• Concave w.r.t. input prices

• Non-increasing VC w.r.t. capital stock

Estimated• Labor cost share is

positive; increasing in input prices

• Concavity is satisfied• Increase in capital cost

with increase in capacity; non-increasing VC w.r.t. capital stock not satisfied

• Cobb-Douglas and identical fixed-effects hypothesis: rejected

LR Results

• Marginally increasing in capital stock increases (not decreases, as expected in cost theory) VC

• Interpretations– Positive sign of coeff. of capital stock indicates

excessive amount of capital stock employed by firms

– Incorrect sign of coeff. of capital stock comes from multicollinearity between output and capital stock• More precise as based on empirical analysis• Causes unexplained

LR Results

• Cause of positive sign of coeff. of capital stock: empirical difficulty in defining/calculating capital stock variable

• Lack of data SR cost studies use physical measures reflect max. available production capacity highly correlated with increasing output multicollinearity

• Solution: calculate capital stock using capital inventory methodEstimation results are inconclusive to LR cost

minimization hypothesis

Economies of utilization and scale

• According to results:– Utilization and scale economies exist– If S/M/L companies increase output with

holding capacity fixed, VC increases less than proportionally

– Increase in output without holding capacity fixed increases TC less than proportionally

– Importance of utilization and scale economies increases with sizeEmpirical results confirm economies of scale in

Swiss electric distribution utilities

Conclusions

• Economies of scale exist for S/M/L utilities

• Inconclusive regarding over-capitalization

• Policy implications:– Utilities should operate as local franchised

monopolies– Redesign on economic incentives to promote

optimal behavior– Encourage competition and merger policy

THANK YOU