ARTICLE IN PRESS

Energy Policy 32 (2004) 949960

Technical and economical analysis of an induced demand in the photovoltaic sector

Umberto Ciorbaa,*, Francesco Paulib, Pietro Mennaa,c,1
b a ENEADirezione Studi, L.Tevere Thaon di Revel, 76, 00196, Rome, Italy " Dipartimento di Scienze Economiche e Statistiche, Universita degli Studi di Trieste, Piazzale Europa 1, 34127, Trieste, Italy c ENEADirezione Studi, Centro Ricerche, 80055 Portici, Italy

Abstract We analyse the potential economic impact of a demand of photovoltaic (PV) devices in Morocco, in terms of induced production and job creation. In Morocco a high potential for PV installations exists as proved by the number of national and international deployment initiatives currently underway. In our study, we rst describe the manufacturing process providing as nal output the solar modules and associate the corresponding costs to each step of the process. These costs are needed to determine the technical coefcients of an ad hoc PV sector to be added to the Morocco Leontief matrix. Then, using the inputoutput methodology, we evaluate the economic impact of the production of 5 MWp modules a year under different hypotheses. A production of 5 MWp PV modules, which require an overall investment of h16.3 millions, increases production of h57.6 millions if cells are locally produced and of h22.3 millions when the cells are imported, creating 2570 and 489 jobs, respectively. These results outline the importance of availability of intermediate inputs in local economical context. r 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Photovoltaics-induced demand; Economic impact; Morocco

0. Introduction Due to systems installation and photovoltaic (PV) infrastructure development, an induced large-scale demand of PV systems can produce meaningful socioeconomic implications in terms of induced production and job creation. The implementation of PV manufacturing facilities may stimulate several economic activities helping the setting up of a local industry and inducing a more environmentally sound development in the selected region. The deployment of PV systems in rural areas is also appealing as it might help to satisfy the need of electrication of large regions of developing countries. It is known that each kWh made available in areas not reached by the grid improves the quality of life in terms of larger food availability, lower transportation costs,
*Corresponding author. Tel.: +39-06-36272502; fax: +39-0636272885. E-mail addresses: ciorba@sede.enea.it (U. Ciorba), francesco.pauli@econ.univ.trieste.it (F. Pauli), pietro.menna@cec.eu.int (P. Menna). 1 Currently at the: European Commission, B-1040 Brussels.

better health conditions and longer life expectancy, better possibility for providing instruction training and information. In other words electricity represents an essential means to start an accelerated scenario for growth and development (Menna and Paoli, 1998). After describing the current status of the world PV module production, the future prospects and the technology cost issues, we provide some details on the PV modules production process, useful if a transfer of the technology to the current industrial environment of the considered region should be implemented. In order to allow for the specic features of PV industry in determining its economic impact we detail the module production cost structure: cost of intermediate inputs, labour, other indirect cost. Then we use these gures to build an ad hoc PV sector added to the existing Morocco IO table.

1. World PV module production The world market for PV has been growing, on the average, at 30% annually for the past 5 years. The USA

0301-4215/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0301-4215(03)00018-1

ARTICLE IN PRESS
950
100 90 80 70 60 50 40 30 20 10 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Others a-Si Multi cSi Single cSi

U. Ciorba et al. / Energy Policy 32 (2004) 949960

Fig. 1. Shipments of PV modules by different technologies (years 19912001) (%) (adapted form different sources).

and Japan have been market leaders over much of the history. However, the European market position has improved steadily over the past 5 years. The total world market was close to 400 MW in 2001, more than 40% increase from 2000, for a business value close to h2.5 billion. Japan, that was behind the other regions in the mid1990s, increased its share becoming the rst manufacturer of solar modules in 1999 and conrming this position in the year 2000 and 2001. The current production of PV generators is mostly based on crystalline silicon technology, which uses either the single-crystalline or the multi-crystalline approach. Single-crystalline silicon had been the industry standard in PV markets, covering 49% of the demand in 1997. However, multi-crystalline Si has taken over the top position with a 44% share in 1998 and has increased its share to 50% in 2001. The contribution by the other technologies (i.e. thin lm) appears small (Fig. 1). Further growth in the world market will be strongly inuenced by PV costs and available technologies. Todays cost of crystalline silicon modules could be estimated considering that the current prices ranging from h3.5 to h4 per peak watts includes all costs as well as marketing and management overheads for that product line. It is worth noting that while the module prices have been essentially stable in the last 45 years, the manufacturing costs have been reducing, increasing the prot margins. The cast-ingot multi-crystalline Si option continues to have module costs slightly lower than those of the single crystal.

Within the process we can recognise the main following segments: (a) (b) (c) (d) (e) feedstock preparation; wafer production; solar cells manufacturing; modules production; and system integration and/or adaptation.

2. The modules manufacturing process Conceptually the process ow diagram starts with the production of material feedstock and includes the whole sequence until the commercialisation and installation of the systems.

Steps (a)(d) are represented in Fig. 2. Lower levels in the process streamline correspond to lower technological content and smaller manufacturing units. As a consequence, smaller investments are required. On the opposite, the production level which is needed to ensure protability increases at higher levels of the streamline. The size of the investment and the technology needs for a new enterprise to enter the sector is strongly dependent upon the entry level in the process streamline and the kind of process integration chosen. It is unlikely that vertical integration will cover the whole process from (a) to (e), it is more likely for a facility to be partially integrated from (c) to (e) or (d) to (e). Moreover, it is unlikely for the rst segment, feedstock preparation, to be carried out within PV sector. As known, the PV industry relies on secondary polysilicon, such as the excess or rejected material from electronicgrade polysilicon feedstock production. The so-called polysilicon material is produced by few manufacturers in the world, mostly to serve the electronic industry. Only a share of this production becomes feedstock for the wafer production. In other words, there is not yet such a thing like a dedicated supply for the PV industry, because the overall demand from PV is still low. At the year 2001 with shipments level of about 4000 MW, considering a conversion of 1520 t of polysilicon per MW, it comes out of a demand of 60008000 t of polysilicon, a capacity that may be still limited to make the manufacturing protable. This means that phase (a) is carried out by industries which do not generally rely on

ARTICLE IN PRESS
U. Ciorba et al. / Energy Policy 32 (2004) 949960 951

Wafer production
Ingot growth Shaping & wafering

Cells manufacturing
Etching & texturing Junction formation & coating preparation

Silicon feedstock

Modules production Material preparation

Paneling Interconnection Metallization

Fig. 2. Flow diagram of the industrial process for modules manufacturing, steps (a) to (d) are represented here.

demand from the PV industry alone and are not strictly part of the sector. This phase is excluded from the analysis. We focus our technical analysis on modules assembling, assuming that wafers and solar cells are an input to the production process. This choice is coherent with the size of the investment we believe adequate, which correspond to 5 MW modules production a year.2

3. The energy sector in Morocco The Kingdom of Morocco is located in North Africa. Its population is nearly 28 millions inhabitants. Moroccan economic activities include export-oriented agriculture based on irrigation, mining (especially phosphates), tourism (2 millions tourists per year with an increasing trend), sheries and transformation industries (food industry, textiles). The energy electricity sector is managed by the Ofce ! National de lElectricite (ONE). The production is mostly based on imported fossil fuel (86%). Hydropower account for 14%. Renewable is in progress. In 1997 the electrication rate of Morocco was still less than 25% (Bakri et al., 1998). To promote development of renewable energy, Morocco has created about 20 years ago the Centre de D!veloppement des Energies Renouvelables (CDER). e This centre aims to (i) develop renewable energies uses,
2 Incidentally, it is worth reporting that BP-Solar, the worlds largest solar energy company, in April 2000 opened a PV panel manufacturing plant in Malaysia. BP-Solar (M) Sdn Bhd is a joint venture company between BP-Solar and Projass Enecorp Sdn Bhd incorporated on February 1999 to manufacture solar systems locally. The company is in full support of the Malaysian Government to enhance localisation programs and plans to carry out research in balance of components for solar PV system. The solar manufacturing plant is the rst solar plant in Malaysia, incorporating the equipment and power technology proven in facilities around the world. The plant received an assistance of h275,810 from Malaysian Technology Development Corporation in terms of Technical Acquisition Fund contributed towards the total capital outlay. The plant is designed to assemble about 5 MW of solar cells per annum (four shifts) into nished modules.

(ii) set up national capabilities for equipment testing and quality control, and (iii) develop pilot and demonstration projects. The PV sector counts currently about 4 MW installed and has a high potential for future development. In rural electrication alone it is estimated that at least 200,000 households are to be electried by means of solar home systems. ONE is presently active in implementing such program based on the nancial scheme offered by the Programme dElectrication Rurale Globale (PERG),3 in which ONE contributes to each SHS installed with a subsidy equivalent to the cost of a panel and a battery. The PV sector is supported by several international programs.
The Programme d!lectrication rurale g!n!ral (PERG) is a e e e national program aimed to accelerate the introduction of electricity to the 30,000 un-electried villages (douars) of Morocco. Because it would be very expensive to extend the grid to some of the remote and scattered douars, ONE, the national utility, has set aside a special fund under the PERG, to permit the electrication of these villages by PV energy. In these regions where ONE does not plan to extend the grid in the medium term (10 years), it has agreed to partially pay for an individual PV Solar Home System on each household (cost of a module and a battery of a 50 Wp system). In the last few years, ONE has thus proposed to have good-quality PV systems installed and serviced by professional PV rms with the aim to have the user pay for the extra costs engendered by the system. By contributing towards the capital cost of the installation, ONE aims at insuring the continuous service of solar electricity at a cost that rural users can afford. However, while it would be relatively straightforward for the PV rms to install a PV system and obtain payment from both ONE and the user upon installation, it is a lot more difcult to ensure that the PV rm would service the systems and that the user would pay for this service on a long-term basis. ONE had thus to experiment various schemes in providing its contribution and directing PV rms in installing and servicing the PV systems. Another important nancial scheme is the Programme dapprovisionnement group!e en eau potable des populations rurales (PAGER) e that was conceived to accelerate access to water in rural areas of Morocco and is administered by the Direction g! n! rale de lhydraue e lique (DGH). However, the PAGER does not have a direct incentive for the use of PV solar pumping stations. The reasons seem related to the fact that despite it is clearly demonstrated that the cost of operating a PV water pumping station is low compared to a diesel pump, its initial capital cost, which is borne mainly by the DGH, may sometimes be higher than for a diesel pump.
3

ARTICLE IN PRESS
952 U. Ciorba et al. / Energy Policy 32 (2004) 949960

Wind energy is also gaining ground. Several regions of the country have high wind potential. Two major projects are being implemented: one in the North of Morocco (Tetouan) with a capacity of 50 MW, and a second one in the South (Region of Tarfaya) with a capacity of 200 MW. Microhydro power is also of interest especially in the mountainous areas of the Atlas where perennial water streams exist. CDER is implementing several demonstration projects and ONE is in the process of setting up the basis for a large-scale program.

4. Analysis of the PV modules production costs In the following we rst consider the overall cost to implement a facility for module production having a capacity of 5 MWp per year. This may be considered the minimum protable size for module production. The investments required include the cost of the building and the equipment. Production costs take into account intermediate inputs, labour and auxiliary services. The economic assessment of the manufacturing facility requires the denition of the specic parameters and their adaptation to the local context. Labour cost, for instance, is determined after the labour organisation has been conceived and is clearly strongly related to local labour market. Similarly, electricity cost, building cost, local availability of the equipment and/or machinery are evaluated referring to the national context of Morocco. For this analysis we assume that the modules are made using multi-crystalline silicon cells (150 mm

150 mm) and that each module has 36 cells. The nominal module power is assumed to be 100 Wp so that 5 MWp/ year correspond to a production of 50,000 modules per year. The number of cells handled is 1,895,000, assuming a 95% yield of the process. The equipment needed for each operation and the corresponding investment cost is detailed in Table 1. The total investment is h2,105,000. A large part of the equipment is not available on the local market, so that h1,863,000 (88.5% of the total investment) goes to import. In this calculations we used international market equipment price (except for building price); in a more realistic feasibility analysis prices should be conservatively augmented to allow for damages and inefciencies due to inexperienced workers. In Table 2 we detail the workers needed for each operation and for administration and auxiliary services as well. We assume that qualied workers needed are locally available, which may not be the case, particularly in rural areas. The number of generic workers is higher than the standard for similar plants in Europe, thus allowing for lower labour cost and productivity in Morocco. The labour cost is determined by attributing to each worker a suitable/proper wage (ranging from 10h per day for not specialised workers to 50h per day for the management). We note that more than 60% of the total labour force required is directly involved in the production process. In Table 3 we analyse the costs of modules production, that is, the costs of the inputs for the production process. The nal cost results in h326 for each module. It is worth noting that 77% of the cost is due to the cells, so it is not surprising that their local availability is crucial in determining the economic

Table 1 Equipment needed and corresponding investment cost for a 5 MWp/year PV modules manufacturing facility of located in Morocco (by operation) Operation Equipment/Apparata Units Unit cost (kh) Testing and selection of the cells Tabbing (interconnection of the cells) Strings formation and module interconnection Visual control Lamination and panelling of the modules Framing the module Testing Outdoor testing Solar cells simulator Automatic tabbing Strings former Xenon lamp EVA-sheet cutter Laminator Solar simulator for module Outdoor testing facility Miscellaneous Subtotal General Total 1 3 4 1 2 4 2 1 150 100 150 2 3 60 250 100 Cost (kh) Local 150 300 600 2 6 240 0 500 100 1898 82 1980 125 2105 0 0 0 100 50 0 0 30 100 Imported 100 100 100 0 50 100 100 70 0 Avaliability (%)

Building (500 m2)

125

ARTICLE IN PRESS
U. Ciorba et al. / Energy Policy 32 (2004) 949960 Table 2 Number of workers employed in each phase of production of the 5 MWp PV modules facility Operation Testing and selection of the cells Tabbing (interconnection of the cells) Strings formation and module interconnection Visual control Lamination and panelling of the modules Framing the module Testing Outdoor testing Personnel directly employed in the process Maintenance and workshop (2/shift) Supervisor (1/shift+ 1 in stand-by) Administrative assistants Accountant R&D engineers Management Director Responsible commercial Responsible administration Secretaries Personnel not directly employed in the process Total Equipment/apparata Solar cells simulator Automatic tabbing for 150 150 mm2 cells Strings former Xenon lamp EVA-sheet cutter Laminator Solar simulator for module Outdoor testing facility Equip 1 3 4 1 2 4 2 1 Workers/shift 1 1 1 2 1 1 3 1 1 Total workers 3 9 12 6 6 12 9 6 2 65 6 4 2 1 2 1 1 1 4 22 87 953

impact of production. On the other hand, the local expenditure for implementing the facility (h244,000) appears to be almost negligible when compared to the production costs.

5. Methodologies for economic impact assessment Inputoutput techniques rst developed by Leontief (1951) are a standard and well-known tool for assessing economic impact.4 The main advantage of the IO analysis is that it provides impact assessment at a sectoral level with a detail which is generally lost in macroeconomic models. The method is substantially based on the hypotheses of linear relationship (with xed coefcients) between inputs and production, which is reasonable for intermediate inputs in the short period (less reasonable for capital and labour). However, these simple hypotheses are also the target of the main criticism on this method, especially in assessing employment effects where the actual relation between production and labour can hardly be assumed to be linear. An alternative approach in assessing employment effects evaluates direct employment impact by making precise assumptions on the industrial organisation of the production (Hofreither et al., 1998) either based on technical considerations or, for existing productions, on information coming directly from the industrial sector
4 An annotated bibliography on IO method can be found in Lahiri (2000).

(Hansen et al., 1998). Indirect employment effect can be calculated by means of multipliers which may come from the intersectoral table again, but also from econometric models. Keynesian multiplier analysis is also widely used, if sectoral disaggregation can be disregarded, and we fully rely on underlying macroeconomic theory. An impact analysis of subsidies effects has been performed for Austria using a non IO technique (Hofreither et al., 1998). The authors based their work on a PV module demand regression model, estimated on data from the past, which includes a subsidies experience. The forecasted installed capacity for the period 19972010 is used to estimate induced production and job creation. Demand of PV can then be translated into direct employment knowing the number of man-years needed to produce, install and provide after sales service of the systems. Indirect effects are accounted for assuming an indirect employment multiplier of 0.45. National income effect is calculated analogously. The assessment of the direct employment creation that incorporates also future developments of the labour productivity is very detailed. On the contrary, the indirect job creation, evaluated by using an exogenous multiplier coefcient, appears not well specied. However, both direct and indirect effects are evaluated only at an aggregate level. The impact of PV systems implementation has been widely studied, in particular for the United States, due to the activated initiatives to incentive PV (one million roofs project). For the sake of comparison, we present

954

Table 3 Direct and indirect manufacturing costs for 5 MWp Si PV modules and their subdivision according to Morocco IO table sectors (kh) Cost (kh) Direct cost Specic cost % Electric and Chemical Rubber and Metal Electricity and Other Insurances Value Total (of module cost) electronic tradable Added products products plastic products water services (h/Wp) 76.7 9.2 2.5 1.3 1.4 1.1 0.5 1.3 1.2 0.5 0.1 0.2 0.7 0.9 0.2 0.7 0.1 89.2 8.9 1.8 0.1 10.8 100.0 12,887 410 614 204 86 291 291 872 297 10 1453 297 10 12,500 410 204 234 176 83 204 191 75 11 28 114 306 12,500 410 204 234 176 83 204 191 75 11 28 114 306

Cells (N.1895000) 12,500 2.500 Consumables 1500 0.300 Glass 410 0.082 Tedlar 204 0.041 EVA 234 0.047 Junction box 176 0.035 Diodes 83 0.017 Tabbling 204 0.041 Other 191 0.038 Electricity 75 0.015 Water 11 0.002 Waste disposal 28 0.006 Spare parts 114 0.023 Labour cost Workers (65 220 10) 143 0.029 Specialised workers(8 220 20) 35 0.007 Management (10 220 50) 110 0.022 Secretaries (4 220 20) 18 0.004 Total direct costs 14,534 2.907 1453 297 10 1760 0.291 0.059 0.002 0.352

U. Ciorba et al. / Energy Policy 32 (2004) 949960

ARTICLE IN PRESS

Indirect cost Maint., ins., services, overheadsa Equipment depreciationb Building depreciationc Total indirect costs Total cost
a

16,294 3.259

319

291

1485

16,294

10% of the manufacturing costs. 15% (per year) of the equipment costs. c 8% of the building cost.
b

ARTICLE IN PRESS
U. Ciorba et al. / Energy Policy 32 (2004) 949960 955

some results from the existing literature. The State of Maryland project aimed to incentive grid-connected residential PV by nancially supporting residential customers of PV systems has been discussed by Cook (1998). The author states that a $4.8 millions cumulative investment will produce returns for over $100 millions for the State of Maryland. This high-level return comes (also) from the fact that while Maryland is a PV module producer, it is an electricity importing state and so money spent on electricity leave the state while that spent on PV systems activates local economy. These considerations conrm the importance of keeping into account local conditions. According to a 1992 estimate, a 10 MWp a year PV module manufacture in San Francisco bay area generates $40 million in direct annual sales, which leads to $55 million in global impact on the regional economy and provides employment of approximately 80 individuals (US DOE, 1992).

We consider that cells can be produced domestically (Case L) or imported as intermediate goods from abroad (Case I). We calculate different technical coefcients for the sector according to these hypotheses, so that two alternative columns of coefcients are added to the Leontief matrix. The impact of the investment for the implementation of the facility is determined separately. The production induced by the investment is to be added to the economic activity induced by modules manufacture to obtain the total impact of the operation during the rst year. As anticipated, the role of the investment is negligible since the bulk of the expenditure is made abroad.

7. Results The h16.3 M investment in the PV sector induces, directly and indirectly, an increase of production that, in turn, induces an increment of required labour inputs. These increments change according to the investigated options outlined above. In the following sections estimated induced production and job creation are illustrated. 7.1. Impact on production As expected, the impact of production is larger when PV cells are produced domestically (Table 4). In this case induced production amounts to h57.6 million, corresponding to an increase of h17 millions of GDP. If cells are imported the impact is less relevant (h22.3 million production), and the GDP increase by h3.9 million. Induced production and value added is detailed for the most relevant sectors in Table 4 and Fig. 3. The investment in the PV sector stimulates the production of the PV sector itself and other sectors (particularly industrial sectors) as well. Demand for PV cells greatly stimulates, directly and indirectly, the production of the electronic and electric parts. This sector is most sensitive to the option chosen (with an induced production in Case L being more than ten times greater than in Case I). Induced production coefcients can be evaluated considering the amount of expenditure needed to build the modules. Coefcients vary considerably from 1.4 (Case I) to 3.5 (Case L) 7.2. Impact on employment Due to the lack of detailed information on occupation in Morocco, the estimation of induced occupation has been performed using aggregate sectors. The Statistical Directorate of Morocco provides data on value added (at current prices) for 1999,

6. Inputoutput analysis for Morocco As already anticipated, we perform our analysis by inputoutput (IO) techniques using the 1990 Morocco IO table with 33 sectors. In IO analysis no change in the technological process can be accounted for. Each column of the IO table represents the amount of inputs required from each sector to produce the total output of the column sector. The impact of any change in nal demand is projected assuming that the shares of cost for each input are kept constant in the short period. This feature of the method may limit the comprehensiveness of the results, particularly when dealing with new technologies, whose technical coefcients are likely to be signicantly different from those of traditional sectors of production. Because production costs of the modules are available, we decided to build an ad hoc PV sector to be added in the IO table so that specic module manufacturing inputs are accounted for. In doing so, we assume that the PV sector does not provide any intermediate goods to the other sectors and that the production is completely absorbed by nal consumption. This means that the row of the modied IO table corresponding to the PV sector is null except for the cell corresponding to Final consumption whose value is equal to the total value of production of the sector. The breakdown of the modules manufacturing cost provides the technical coefcients of production for the ad hoc PV sector. The costs analysed in the previous section have been reallocated by economic sector (see Table 3) assuming all inputs of production but cells are locally available.

ARTICLE IN PRESS
956 U. Ciorba et al. / Energy Policy 32 (2004) 949960 Table 4 Direct production (kh), total induced production (direct and indirect) (kh) and job creation (units) for most relevant sectors due to a h16.3 millions demand in PV sector Distributed Production (kh) L 1 22 18 24 32 7 8 6 28 25 19 20 31 Photovoltaic Electric and electronic products Basic Metal Industries Chemical products Other tradable services Rened oil Electricity and water Solid fuels and crude oil Trade and Transport Rubber and plastic Metal products Manufacture of equipment Insurance Other Total 16,294 16,350 4477 3718 2326 1612 1492 1248 1244 1210 1082 599 539 5389 57,581 I 16,294 501 323 1039 648 271 335 205 195 656 301 44 500 947 22,259 Value Added (kh) L 1485 5567 1573 1138 1492 315 377 975 709 400 406 248 169 2317 17,171 I 1485 170 113 318 415 53 85 160 111 217 113 18 157 446 3861 Job creation L 87 791 223 162 262 45 53 138 132 57 58 35 30 497 2570 I 87 24 16 45 73 8 12 23 21 31 16 3 28 103 489

32

Other Other tradable activities Insurance Trade and Transport Rubber and plastic Chemical products Electric and elctronic products Manifacture of equipment Metal products Base Metal Industries Electricity and water Refined oil Solid fuel and crude oils
0 1000 2000 3000 4000 5000 6000 16,350

8 18 19 20 22 24 25 28 31

I L

Fig. 3. Induced production of main sectors (millionh) due to a h16.3 million demand to PV sector.

for each of the main sectors of the economy. Data on number of employees for the same sectors are available for 1999.5
5 The sectors considered are: Primary. Manufacturing industry, Constructions, Commerce, Transport and communications, Services. Primary sector covers almost half of the occupation while it contributes only for 15% of the GDP. Industry represents 28% of GDP and employs 14% of labour force, for services the shares are, respectively, 27% and 17%. The other sectors together cover 11% of GDP and employ 9% of labour force. The overall GDP is h32.6 billions and the total workers are 9,341,600. The GDP per employee is h3500 as a whole, with relevant differences among sectors (from 1000 of Primary sector to 7000 of Industry and Transport). Monetary conversions in Euro have been made using an exchange rate of h0.10159 per Dirham.

By calculating the number of employees per unit of value added and multiplying the ratio for the induced value added in the sectors considered, we obtain an estimate of the occupational impact of the investment, also detailed in Table 4. Induced job creation is 2570 employees concentrated in the electronic, the PV and the metallic sector (Case L). In Case I the total increment of occupation is 489 employees concentrated in the chemistry, the PV and the plastic sector. 7.3. Sensitivity analysis Results of inputoutput analysis are affected by the uncertainty in estimating sectoral intermediate input

ARTICLE IN PRESS
U. Ciorba et al. / Energy Policy 32 (2004) 949960 957

Fig. 4. Probability density functions for simulated estimates of induced production for main sectors in case L assuming a normally distributed error with 5% standard deviation on input costs (thousands h).

costs, which determine technical coefcients. Even if all the coefcients in the table are estimates and contribute to the nal error, in our investigation one of the main sources of uncertainty we are concerned with derives from the determination of input costs of the PV sector. However, this represents the only part of the table for which we are able to quantify the error. To determine how uncertainty on PV costs propagates to the nal estimates, a simple simulation study has been carried out. We assume that the values in nal row of Table 3 are known with an error normally distributed with zero mean and standard deviation of 5%. By independently simulating the value for each cell according to the distribution assumptions above, we generate 5000 independent and identically distributed PV sector columns. Each of these columns in the 33 33 IO matrix provides an input for calculating the production. Should costs be the only source of uncertainty, the output of the simulation would have been a sample from the probability distribution of induced production estimate (estimator). However, in the real context, the simulation allows us to understand how sensitive the nal results are to our hypotheses on PV sector input costs. A gaussian error with 5% standard deviation in input cost leads to nal estimates of induced production with

a standard deviation of 4.7% (Case L) and 5.1% (Case I). This allows us to conclude that our results are reasonably robust with respect to misspecication of the cost structure of module manufacturing. Analogous conclusions are drawn as far as sectoral-induced production is concerned by examining their simulated estimates (Figs. 4 and 5 where histograms of simulated estimates are shown). 7.4. Comparison of IO analysis: Morocco vs. Italy It is worth comparing the results of our inputoutput analysis for Morocco with a similar study made for Italy. Of course, in doing the comparison, we should remind that the structure of the two economies is radically different because of the different development level. From a methodological point of view, the differences between the two economic structures translate in the fact that the IO table available for Italy is far more detailed and uses a classication of the sectors which is not coherent with that used in the Morocco table.6 In doing the comparison, we assumed for Italy the same cost structure as for Morocco. Labour cost,
6

Meaning that it is not simply a renement of the Morocco one.

ARTICLE IN PRESS
958 U. Ciorba et al. / Energy Policy 32 (2004) 949960

Fig. 5. Probability density functions for simulated estimates of induced production for main sectors in case I assuming a normally distributed error with 5% standard deviation on input costs (thousands h).

however, ranging from 17kh/year for unskilled workers to 85kh/year for the management is higher in Italy.7 This means that the manufacturing costs for 5 MWp modules are higher in Italy (17.936 Mh) than in Morocco (16.294 Mh). Because in the context of Italian economy, cells can be locally produced, we will consider only this case in the comparison. The expenditure of 17.936 Mh lead to induced production of 28.240 Mh (excluding PV sector), corresponding to a multiplier of 2.6, signicantly lower than the one obtained from Morocco analysis. Similarly, the number of jobs created is lower in Italy (605) than in Morocco (2570). These results are not surprising. Industrialised and developing countries have different stocks of capital per worker determining different levels of productivity. Even if the initial investment and the production multiplier were the same, job creation should have been lower in Italy because of the higher level of productivity. In fact, between the two countries the difference in productivity is quite relevant. For instance, in the manufacturing sector the value of production per worker is 5 times higher in Italy than in Morocco. If
7 The source of these gures is INPS, the Italian National social security administration.

we consider the whole economy the difference is even higher (13 times).8 As far as the sectoral splitting of induced production is concerned, we have to take into account some difculties in doing the comparison. In fact, the available IO table for Italy is made up of 92 sectors of production while the available Morocco IO table has 33 sectors. The 92 Italian sectors classication is not just a rening of the classication used for Morocco. It means that to make a comparison we need to dene suitable aggregated sectors, that are presented in Table 5. The main difference which can be seen between the two countries is the heavier role of services sector with respect to primary sector and manufacturing sector. This result depends crucially on the higher degree of integration between the services sector and the productive system of industrialised countries. In fact, in recent years, demand of services in industrialised countries grew up to satisfy the demand of services coming from nal consumers and the demand of intermediate services coming from the manufacturing sector. In developing countries the sector provides services mainly to the nal consumers and the share of production used by
8 Sources: ISTAT, Annuario Statistico Italiano 2000; Statistics Directorate, Maroc en chiffres 2001.

ARTICLE IN PRESS
U. Ciorba et al. / Energy Policy 32 (2004) 949960 Table 5 Comparison of economic impact for Italy and Morocco due to a h16.3 millions demand in PV sector Induced production Italy kh Photovoltaic Electric and electronic material Other services Metalmetal products Other manufacture Rubber and plastic Chemicals Trade and transport Electricity and water Coke and oil Insurances Paper and publishing Communications Constructions Wood products Dressing-textiles Food Manufacture of transport equipment Agriculture, hunting, forestry Tobacco and beverages Total (excluding PV) Total (including PV) 17,936 14,592 3172 2605 2605 1343 893 745 533 371 361 355 187 153 105 97 43 37 30 12 28,240 46,176 % 51.67 11.23 9.22 9.22 4.76 3.16 2.64 1.89 1.31 1.28 1.26 0.66 0.54 0.37 0.35 0.15 0.13 0.11 0.04 Morrocco Kh 16,294 16,350 2504 5559 4476 1210 3718 1244 1492 1248 539 820 178 83 818 236 176 333 274 29 41,287 57,581 % 39.60 6.06 13.46 10.84 2.93 9.01 3.01 3.61 3.02 1.31 1.99 0.43 0.20 1.98 0.57 0.43 0.81 0.66 0.07 Job creation Italy units 87 163 105 49 49 21 12 66 8 1 2 9 10 7 5 4 1 1 6 0 518 605 % 31.57 20.29 9.49 9.49 3.97 2.30 12.69 1.45 0.14 0.42 1.66 1.88 1.41 0.90 0.76 0.19 0.20 1.16 0.03 Morrocco units 87 791 290 281 240 57 162 132 53 138 30 25 24 12 32 14 4 13 183 3 2483 2570 % 31.84 11.67 11.32 9.67 2.29 6.51 5.33 2.15 5.57 1.20 1.00 0.98 0.49 1.27 0.57 0.14 0.51 7.36 0.12 959

intermediate consumption is negligible if compared with industrialised countries.9

8. Final discussion The study we considered requires a technology transfer to support the development of countries which have, for historical reasons, an economic level not sufcient to attain the standard of living potentially achievable through more advanced technologies. To be performed with effectiveness, the transfer requires a clear knowledge of the needs as well as the potential of the beneciary. To this concern, the role of governments should be devoted to understand local conditions and to create the background for private investments in the country. This is in order to amplify the effect of public interventions, they being generally insufcient to satisfy the needs of the recipient country. The background for private investments is essentially represented by the institutional framework able to assure some kind of return for the investment as well as those information, knowledge and ability which are conditions to make the investment a benet not only for the enterprise carrying it but also for the developing country.
9 About integration between services and manufacturing sector see Momigliano and Siniscalco (1982).

It has been made clear, for instance, to what extent local availability of cells brings to very different returns of the investment in the local economy, the induced production decreases the expenditure from 3.5 times to 1.4 times (when 77% goes abroad to import cells). The same considerations apply to the expenditure needed to implement the facility which has a negligible effect on local economy, because only a limited and low technology contents part of goods required are domestically available. For a more relevant demand the potential negative impact on external trade balance should be allowed for. In fact, the expansive effect might be partially displaced by restrictive policies adopted in order to improve the external trade balance. This kind of effects can be very important when the increment of the production is obtained in technologically advanced sectors that are heavily dependent on the import of components from developed countries. It should be reminded that the analysis we performed does not consider all economic issues involved. Introducing PV systems in Morocco brings direct benets to users in rural areas not yet electried, contributing to the improvement of life conditions. The access to electricity services brings to time savings and day extension, which may incentive productive activities. The new situation might also improve access to information and demand of technological products (like appliances), thus creating indirect economic effects

ARTICLE IN PRESS
960 U. Ciorba et al. / Energy Policy 32 (2004) 949960

difcult to quantify (Mulugetta et al., 2000; van Campen et al., 2000). PV installations should be considered as a rst step of a developing path toward industrialisation of rural areas. Areas rst electried by PV will be likely to be grid connected once they reach a signicant electricity demand. At that point, PV installations are likely to become part of the local support for the grid. Moreover, economic aspects are not exhaustive in assessing this kind of project. Attention should also be paid to the potential, from the point of view of environment saving (Kim et al., 2000) of the transferred technologies. It must be recognised that the new technology which is introduced in the developing country will replace probably an old existing one, or some alternative technology used for example in the past in the industrial countries during their developing phase. The impact of these alternatives on the environment, if developed without control, represents what we are saving through the transfer. In our example, we can roughly estimate that the choice of PV, which is a zeroemission source, might save, with respect to traditional technologies, 0.56 t CO2/MWh.10

in the more realistic case that cells need to be imported, most of the expenditure will not advantage local economy. On the other hand, a more integrated process incorporating the cells manufacturing step would require the specic technological know-how and the proper job-training, both likely not to be locally available and need to be transferred.

References
Bakri, M., Fassi-Fihri, A., Simonis, P., 1998. Change of scale in the implementation of decentralized rural electrication: the case of the German cooperation in Morocco. Proceedings of the Second World Conference on Photovoltaic Solar Energy Conversion, Vienna, 610 July, pp. 34103412. van Campen, B., Guidi, D., Best, G., 2000. Solar Photovoltaics for Sustainable Agriculture and Rural Development. Environment and Natural Resources. WP No. 2. FAO, Rome. Cook, C., 1998. The Maryland solar roofs program: state and industry partnership for PV residential commercial viability using the state procurement process. Proceedings of the Second World Conference on Photovoltaic Solar Energy Conversion, Vienna, 610 July, pp. 34253428. Hansen, R.E., Mcillivray, R.G., McHugh, W.M., 1998. An emerging center of gravity for Washington State: an analysis of economic and scal impacts of high-technology manufacturing facilities. Columbia River Economic Development Council. Hofreither, M.F., Rath, S., Rimpler, G., 1998. Economic effects of subsidies for grid coupled PV systems in Austria. Proceedings of the Second World Conference on Photovoltaic Solar Energy Conversion, Vienna, 610 July, pp. 33843387. Kim, S.H, MacCracken, C., Edmonds, J., 2000. Solar Energy Technologies and Stabilizing Atmospheric CO2 Concentrations, Progress in Photovoltaics: Research and Applications, Vol. 8. Wiley Chichester, UK, pp. 315. Lahiri, S., 2000. Professor Wassily W. Leontief, 19051999. The Economic Journal 110, F695F707. Leontief, W., 1951. The Structure of the American Economy 1919 1939. Oxford University Press, New York. Menna, P., Paoli, P., 1998. Photovoltaic Installations for the Electrication of Rural Southern Mediterranean Regions, Progress in Photovoltaics: Research and Applications, Vol. 6. Wiley, Chichester, UK, pp. 357363. Momigliano, F., Siniscalco D., 1982. Note in tema di terziarizzazione e deindustrializzazione, Moneta e Credito no. 138 . Mulugetta, Y., Nhete, T., Jackson, T., 2000. Photovoltaics in Zimbabwe: lessons from the GEF solar project. Energy Policy 28, 10691080. US DOE, 1992. Economic impact of a photovoltaic module manufacturing facility.

9. Conclusions We carried out a technical analysis of an industrial initiative in the Photovoltaic sector and evaluated the economic result of a demand of PV devices in terms of gross domestic production and job creation. To perform the analysis we build an ad hoc IO table to allow for the specic industry considered. Because the results of the analysis are affected by the uncertainty in estimating technical coefcients of PV sector, a sensitivity analysis has also been performed. The simulation allows us to conclude that impact analysis results are reasonably robust with respect to misspecication of the cost structure. We also discussed the results of a similar analysis carried out for Italy, underlying that, because of the radical differences among the two economies, the comparison cannot be stretched too much, not least for methodological reasons. The relevant role played by the local availability of input factors like solar cells is clearly seen in our results:

10 Ratio of CO2 emissions from electricity generation and total electricity production for 1997 in Morocco as recorded by IEA (CO2 emissions from fuel combustion 19711997 and Energy Statistics for Non-OECD Countries 1998).