Coverage Probability and Achievable Rate Analysis of FFR-Aided Multi-User OFDM-Based MIMO and SIMO Systems
Summary (5 min read)
1 Introduction
- Oil and gas prices have been fluctuating over the past two decades, but started declining in the 2010s.
- In a 2014 report, the European Commission finds that energy retail prices have increased by 4% annually across all member states over the 2008-2012 period,1 and the average increase in electricity retail prices between 2008 and 2013 amounts to 28%.2.
- This paper provides an empirical analysis of cost pass-through in the German retail market for electricity.
- The independent firms, which the authors assume to be most competitive, exhibit 15-20% higher pass-through rates to the competitive market segment; the pass-through rates to baseline tariffs do not significantly differ across firms.
- These dimensions of heterogeneity are key to go beyond the estimation of average pass-through rates and thus understand the sources of pass-through.
2 Literature
- The literature on pass-through is quite extensive.
- There is also a literature using reduced form approaches.
- Deltas (2008) studies asymmetric pass-through in the US retail gasoline market and finds prices respond faster to wholesale price increases than decreases.
- This asymmetric response, as well as the speed of adjustment, are shown to be a consequence of retail market power.
- The authors study is therefore the first to estimate cost pass-through to electricity retail prices using a large and disaggregated panel dataset including both price and cost data, as well as distinguishing several dimensions of heterogeneity in pass-through rates.
3 The German electricity market
- The German market is characterized by a vertical structure comprising a generation segment, a wholesale market, and retail markets (see figure 1).
- The transmission network ensures that energy generated or imported is delivered to regional supply companies, which then distribute it via low or medium voltage distribution networks to energy re- tailers and final customers.
- Finally, a parallel balancing market ensures that the necessary voltage is maintained in the network at any given time.
Insert Figure 1 here
- The generation segment in Germany is dominated by three vertically integrated, although legally unbundled, utilities: E.ON, RWE, and Energie Baden-Württemberg (EnBW).
- They jointly meet 2/3 - 3/4 of the total German electricity demand.
- Other companies, including EWE and RheinEnergie, collectively represent [55-65 per cent] of this market."the authors.
- All big players lost some market share over time, yet, at the national level, they continue to cover almost half of the market.
- This rather aggregated picture is partially misleading.
Insert Figure 2 here
- While several retailers offer different tariffs in each of these regions, incumbent providers are legally obliged to sell energy at a baseline tariff to all household customers who do not explicitly choose another provider.
- Accordingly, this baseline tariff constitutes an upper bound for the energy retail prices in a given region because it is automatically chosen by customers unwilling or lacking the information to switch supplier.
- The number of 8According to German law, the incumbent is the firm that serves the majority of household costumers in a local market at a given point in time.
- The incumbent provider is newly defined every three years.
- Households switching providers has grown at an increasing rate over time, yet incumbent providers have maintained a very strong customer base.
3.1 Retail price structure
- The authors empirical analysis focuses on the evolution of retail prices, in particular on their relationship with wholesale prices and network charges.
- On the one hand, they are affected by electricity wholesale prices that constitute the main essential input for retailers.
- On the other hand, they are also strongly influenced by other factors, including the cost of transmission and distribution, concession fees, as well as taxes and other fees.
- In its 2012 monitoring report (Bundesnetzagentur (2012)), the German regulator discusses the structure of retail tariffs in depth for household customers, whose national average composition for the 2006-2012 period is reported in figure 3.
Insert Figure 3 here
- These average values are useful to understand the various components of retail tariffs.
- Therefore, retail tariffs present a lot of cross-sectional variation across, as well as timeseries variation within regions.
- The German regulator reports that the cost of energy purchase varies within different types of firms.
- Since 2007, entrants achieved on average more favorable conditions mostly because they buy energy from the wholesale markets through shorter-term contracts and wholesale energy prices have decreased.
- Customers living in urban areas tend to switch more because they tend to be better informed and because they face a larger set of available tariffs.
4 The data
- The main data source for the analysis is the price comparison site Verivox, which provides highly disaggregated data on energy retail prices, specifically, monthly price data between January 2007 and August 2014 for 8,192 different postal codes (located in 6,205 cities across all 16 German states) from 893 different incumbent providers and 497 different non-incumbent providers.
- In Table 1, the authors present summary statistics on retail prices in the data-set.
- These bounds are used to define the variable ’price dispersion’ which represents the difference between the most and least expensive tariff in each postal code and period.
- Thus, e67.1 per mWh or more than 25% could have been saved by switching from the baseline tariff to the least expensive tariff.
- Municipal providers make up 19% of incumbents, while another 19% have a joint ownership structure.
Insert Table 2 here
- The data on the costs of purchasing and transmitting electricity are obtained from EEX and ene‘t respectively.
- The authors aggregate these cost factors (network charges, concession fees and wholesale energy) into a single cost variable, indicating the per-mWh cost of providing energy.
- Note that while network charges and concessions fees are postal code-specific, thus varying across regions and time, wholesale prices are uniform across Germany and only vary over time.
- 9All their findings are robust to using month-ahead, quarter-ahead or half-year ahead wholesale prices instead.
- Finally, looking at the evolution of costs over time (figure 4, panel (c)) the authors see a significant peak in 2008, while costs mostly remain in the range of e 110 to e 120 per mWh in the years before and after.
Insert Figure 4 here
- As discussed above, the heterogeneity in costs, demand, and competitive conditions at the regional level leads to significant retail price dispersion across local markets.
- Figure 5 shows this geographical dispersion for one specific tariff – baseline tariff for a consumption of 2,800 kWh – for one particular point in time – the year 2010.
- The different colors represent the quartiles of the price distribution.
Insert Figure 5 here
- The authors observe significant differences in the level of the baseline tariffs across regions.
- Baseline tariffs are highest in the north-eastern part of Germany, where the price of consuming one mWh of electricity lies in the (257, 302] interval in almost all regions.
- In the southeastern part, this cost range is substantially lower with almost all regions belonging to the [207, 248] interval.
- The west of Germany has more homogeneous prices, with most values lying in the second and third quartile.
- The different colors represent the quartiles of the price-dispersion distribution.
Insert Figure 6 here
- There is significant price dispersion in each area and significant cross-area differences in the size of this dispersion.
- It varies between e26 (lowest value of the first quartile) and e107 (highest value of the fourth quartile) by mWh consumed.
- This is quite substantial given that the average baseline tariff (best) price for consumption of one mWh is around e260 (e190).
Insert Table 3 here
- Finally, the authors employ a large number of control variables at the postal code level which are also obtained from ene‘t.
- The total population, the number of available distribution grids, their total length, the capacity of energy transformers, the total number of household connections (metering points), network losses in percent, cost of network losses in e, as well as total energy transmitted, also known as These include.
- Table 3 contains summary statistics on network charges, wholesale prices and the control variables.
5 Model and estimation equation
- The empirical model the authors apply to the data aims at estimating the pass-through rates of network charges and wholesale prices on retail tariffs, while at the same time controlling for local supply and demand conditions.
- Even though electricity is a rather homogeneous good, contracts are perceived by costumer to be vertically differentiated, as the several tariffs of the different retailers in a given regional market are offered under different conditions (length of the contract, conventionally produced or ’green’ electricity, bonuses, quality of service, etc.).
- The authors also add a large set of fixed-effects.
- In a first step, the authors assume that β is common to all observations, thus reflecting the average pass-through of costs to retail prices in the whole sample.
- Third, the authors estimate firm-type f specific pass-through rates so that they obtain separate pass-through rates for municipal, big-four, independent and other retailers as well as the different tariff types βi f .
6 Results
- The authors discuss the main regression results.
- In the first column, the authors look at the most aggregate specification where the tariffs are pooled.
- Tariffs are positively related with demand drivers such as population and the number of connections as well as with costs drivers such as network loss and their costs.
- 12This level of clustering aims at capturing that many providers – especially the big four – mostly offer tariffs that are homogenous across regions.
- With measures of efficiency such as the number of grids, the total grid lengths, and the transformer capacity.
Insert Table 4 here
- The authors then split the sample according to the two tariff types – ’incumbent base’ and ’overall best’– which allows us to estimate heterogeneous pass-through rates depending on the customers’ types.
- The authors expect customer who buy the ’best’ tariff to be better informed and to have smaller switching costs.
- Therefore, the authors expect these tariffs to be more competitive and to reflect more strongly changes in costs.
- The coefficient estimate of the cost variable increases to 70% for the overall best tariff, while it drops to 49% for the baseline tariff.
- In the next step the authors allow the pass-through to vary across firm types in an effort to analyze whether firms differ with respect to downstream market power.
Insert Table 5 here
- The rate of cost pass-through to the incumbent base tariff is close to and not significantly different from 49% for all firm types, which is the same pass-through rate the authors found in the pooled specification (table 4, column 2).
- In the more competitive market segment, the authors find that independent firms exhibit the highest degree of pass-through.
- The next dimension of heterogeneity that the authors exploit in the econometric analysis is time.
- In table 6 the authors report the results for the specifications where they estimate a time-dependent pass-through for the different tariffs.
- In contrast, the best tariff passthrough rate, starts out at around 40% at the beginning of the sample period, increases in 2010 and, after a dip in 2011, becomes almost unity for the final years in the sample.
Insert Table 6 here
- This result appears particularly interesting as it suggests that, after 2011, the passthrough rate for the most competitive part of the market, i.e., the best tariffs, is almost complete.
- This is consistent with almost perfect competitive outcomes in this segment of the market.
- The final step of their empirical analysis is to allow the maximum amount of heterogeneity in the pass-through rates, which are estimated to be tariff, firm, and time-specific.
- Because of the large amount of estimated coefficients, the authors present the results graphically in figures 7 and 8 for the baseline and best tariffs respectively.
Insert Figure 8 here
- While the authors see a different evolution of pass-through rates across tariffs, i.e., market segments, as before, they do not find significant differences across firms.
- Thus, all types of firms in their data exhibit remarkably similar dynamics in their pass-through behavior over time.
- For most time periods, the authors see that the pass-through rates of independent firms tend to be the highest, but not by a large margin.
- Again, these results seem to suggest that the pass-through rate is mostly driven by consumer behavior as represented by the different tariff types rather than by firms’ characteristics.
7 Conclusion
- In this paper the authors study the pass-through of cost shocks to household retail electricity tariffs in Germany.
- The authors have precise information on the two major cost drivers for electricity retail prices – the regulated network fee and the wholesale electricity prices – which together constitute more than 2/3 of the cost of providing electricity to household customers and are able to control for most other cost factors through several time-varying drivers and numerous fixed effects.
- They are significantly larger for those segments of the markets where demand is more elastic because consumers have lower switching costs and consider products to be less differentiated, while they are higher in market segments where the opposite is true.
- The average pass-through of 60% decreases to around 50% for the incumbents’ baseline tariffs and increases to 70% for tariffs designed for the more mobile costumers.
- The differences across different firm types appear to be limited: while the changes over time are substantial, the pass-through rates of different firm types tend to move in tandem and are not significantly different from each other.
Did you find this useful? Give us your feedback
Citations
31 citations
Cites methods from "Coverage Probability and Achievable..."
...In conventional performance analysis of cellular networks, the regular hexagon cellular coverage model has been widely adopted in [7] and [8]....
[...]
...Utilizing the fractional frequency reuse scheme, the regular hexagon cellular coverage region was divided into the center region and the edge region to investigate the coverage probability and the available rate for cellular networks [8]....
[...]
24 citations
Cites methods from "Coverage Probability and Achievable..."
...The rate of FFR is given by [24], [32]...
[...]
...Similar to the [32], users with SIR higher than St over 25 PRBs or more than 25 PRBs are classified as cell-centre users, otherwise they are classified as cell-edge users....
[...]
18 citations
Cites background or methods from "Coverage Probability and Achievable..."
...Some early work by the authors [17] reveal that the combination of FFR with MU-MIMO not only provides better coverage when compared to a MU-MIMO system without FFR, but also yields nearly 22....
[...]
...The average rate R can be written as [17]...
[...]
...An FFR-aided distributed antenna system (DAS), an FFR-aided picocell, and an FFR-aided multi-user MIMO was studied in [15]–[17], respectively....
[...]
10 citations
10 citations
References
45 citations
"Coverage Probability and Achievable..." refers methods in this paper
...To calculate this pdf, we model the cell shape by an inner circle within a hexagonal cell [33], and assume that the users are uniformly distributed....
[...]
44 citations
41 citations
39 citations
"Coverage Probability and Achievable..." refers background in this paper
...The optimal configuration of FFR is determined in [17] for a high-density wireless cellular network....
[...]
28 citations
"Coverage Probability and Achievable..." refers background or methods in this paper
...An analytical framework of calculating both the coverage probability (CPr) and the average rate of FFR schemes was presented in [10] and [11] for homogeneous single input single output (SISO) and MIMO heterogeneous networks, respectively, using a Poisson point process (PPP)....
[...]
...However, the authors of [10], [11] assumed having an unplanned FFR network, where the cells using the same frequency set are randomly allocated....
[...]
...Hence, two cells using the same frequency for the cell-edge users may in fact be co-located [10], [11]....
[...]
...FFR schemes have been lavishly studied using both system level simulations and theoretical analysis [6]–[11]....
[...]