1. Summary and decisions asked from the team

The idea to test is: today Java-Bali has excess generation capacity while the outer islands add new coal plants; if the Java grid were connected to the outer islands, would national CO₂ emissions fall? Seven points frame the discussion.

• The policy basis is real but the window is narrow. PLN's own plan (RUPTL 2025–2034) has the Java-Bali reserve margin falling from 44.2% in 2025 to 35.7% in 2028 and 33.1% in 2034 ^[C1]. Above a 35% standard the surplus is 3.2 GW in 2025, 1.0 GW in 2027 and zero from 2030; above a 30% standard it stays at 1.5–2.1 GW from 2030 (our calculation) ^[C3].

• The planned big link runs the other way. The Sumatera–Java link (ISJ, COD 2031) is designed to move hydro and geothermal power into Java, 1.7 GW in 2031 and 2.6 GW by 2033 ^[C4]. Only the Java–Bali 500 kV cable (2030) and possible Bali–Lombok–Sumbawa links point the way our idea needs ^[C5].

• The outer-island shortage is mostly not in the plan. The five largest outer systems are planned with reserve margins of 26–52% ^[C7]. New coal in the plan is 3.7 GW of mine-mouth units ^[C8], while captive coal for smelters, which sits outside the plan, was 19.3 GW in 2025 ^[C10].

• IndoTERM-E does not see surplus or shortage. In the database every technology has the same cost in every region, the base-year electricity flows between regions are not physical (41 TWh from outer regions into Java-Bali against 11 TWh the other way) ^[C16], captive coal is almost absent, and the sourcing elasticity for electricity is 1.0 ^[C17]. The situation has to be built into the scenario by design.

• The literature gives a conditional answer. A link cuts CO₂ only if the generation it displaces is dirtier than Java's marginal supply plus line losses. Official 2019 grid factors range from 0.80 t/MWh (Jamali) to 1.1–1.7 (Kalimantan grids) ^[C19]. No study tests our exact question ^[C21].

• Recommendation. Use the dynamic model for the headline result, built on the plan (baseline plan against an alternative plan, as TERM-Brazil did), after a short static stage that debugs the levers and calibrates the link. Present the result as a map of conditions, not a single number.

• A possible home in GMMI. The Round 2 specification of the model intercomparison names investment in grid capacity as an example of green spending and invites a Modeler's Choice scenario ^[C27]. This idea could be our contribution.

2. The policy question and the evidence

2.1 The idea, as the team framed it

The baseline is a Java that has more electricity supply than it needs, and outer islands whose shortage is closed by building new coal plants. The policy is to connect the Java grid to the outer islands and see what happens to emissions. The team note adds four policy questions: the change in combined emissions and the conditions for it to be negative; the effect on the cost of supply (BPP), subsidy and PLN finances; who gains or loses by region and household; and whether the idea brings Indonesia closer to the 2040 coal phase-out pledge or prolongs Java's coal fleet.

2.2 How large is the Java-Bali surplus, and for how long?

RUPTL 2025–2034 (Table 5.86) reports the reserve margin of the Java-Bali system as the net capable capacity minus the net peak load ^[C1,C2]. Whether this is a surplus depends on the standard. PLN planning documents use a minimum of 30% (stated for Sulbagsel in the plan itself ^[C7]) and the team note cites a 35% PLN standard. Figure 1 shows both.

Three readings follow. First, the surplus is largest now and shrinks as demand grows about 4% a year while Java adds little coal capacity. Second, at the 35% standard the surplus is gone by 2030; at the 30% standard about 1.5–2.1 GW remain, which is the same order as the planned 2.6 GW link. Third, capacity is not energy: dividing the plan's 2025 energy by fuel (Table 5.143) by the capacity in Table 5.86 gives about 32% for Java's 14 GW of gas plants and about 77% for its 27.5 GW of coal plants (the team note, using other capacity figures, gets 30% and 72%) ^[C29]. The unused room is mostly in gas. If so, power exported from Java need not come from coal, which matters for the emission sign (section 5.3). The plan also has Java's gas generation rising 2.4 times between 2025 and 2034 (pdf p483), so this room is temporary too. ESDM figures reported in the press give a surplus of about 7 GW at end-2022 and 4 GW at end-2023 ^[C12]; these are secondary and use another basis.

2.3 What does the plan say about the outer islands?

On paper no large outer system is short. Table 1 lists planned reserve margins by system. The shortages reported in the field come from late projects, plant outages and grid faults rather than from missing capacity in the plan. The plan itself attributes the Sulawesi shortage to late projects (PLN and IPP), seasonal loss of hydro capacity and the non-execution of the plant relocation from Java-Bali planned in RUPTL 2021–2030 ^[C28]. It adds that the relocation of a Java combined-cycle plant did not happen because Java needs those units to manage the entry of intermittent renewables ^[C28], which qualifies the idea that Java's idle gas plants are free to supply others. The generation cost is much higher outside Java (Table 2).

Table 1. Planned reserve margin (%), RUPTL 2025–2034, RE Base scenario ^[C7]

Reserve margin = net capable capacity less net peak load, as a share of net peak load. Outer-system peak load grows by 12.4 GW over 2025–2034 in these five systems, against 14.8 GW in Java-Bali. Maluku, Papua and Nusa Tenggara systems are not in this table.

Table 2. PLN's estimated average generation cost (BPP pembangkitan), RE Base, 2025 ^[C13]

Rows 'Estimasi BPP Pembangkitan Rata-rata' of the power balance tables, pdf pages 354, 338, 374, 396, 386, 365, 405, 411, 417.

2.4 Interconnections in the plan and their direction

Table 3. Interconnections and their direction ^[C4,C5,C6]

The ISJ is justified in the plan by Java's limited renewable potential: it moves base-load hydro and geothermal from Sumatera to Java (p529). Our idea therefore fits only parts of the plan: the Java–Bali–Nusa Tenggara corridor, a link designed to run in both directions, or an accelerated link built while the surplus lasts.

2.5 Coal outside Java: utility plants and captive plants

The plan adds 52.8 GW in the RE Base scenario (27.4 GW renewables, 6.3 GW coal, 7.6 GW combined-cycle gas, 4.8 GW gas turbines and engines, 6.7 GW storage) and 69.5 GW in the ARED scenario (42.6 GW renewables) ^[C8]. The 6.3 GW of coal is the same in both: 3.7 GW mine-mouth (PLTU MT) and 2.6 GW other, of which 2.0 GW is Suralaya 9–10 in Java. (The team note quotes the 69.5 GW total with RE Base figures; the 69.5 GW belongs to ARED. We suggest correcting the note.) Examples of planned outer-island coal: Jambi-1 (600 MW, 2030, PPA signed), Jambi-2 (600 MW, 2032), Kalselteng 3 (200 MW, 2029) ^[C9]. Cancelling a unit with a signed PPA triggers termination costs.

Captive plants for smelters are a different order of magnitude. CREA and Global Energy Monitor count 19.3 GW operating in 2025, 4.49 GW added in the year to July 2025 (about 80% of all coal additions), concentrated in Central Sulawesi and North Maluku, and a pipeline that could take the total above 31 GW ^[C10]. Captive plants are not on the national grid, are outside the RUPTL, and are excluded from the JETP plan. The 3.7 GW of mine-mouth coal that a Java surplus could displace is one fifth of the captive fleet, and it sits in Sumatera and Kalimantan, away from the eastern corridor. ESDM's own roadmap says that supplying smelters by interconnection needs further study ^[C11].

2.6 What this means for the question

• The surplus window is short (to about 2028 at the 35% standard) and may close before a link is built; a link that runs in both directions is a no-regret investment but changes what the model must show.

• The coal that grows fastest outside Java is captive, which a grid link cannot displace unless the smelters connect.

• The emission effect is therefore a conditional result. We propose to frame the paper as: under which conditions does connecting Java to the outer islands lower CO₂, and by how much?

3. The tools: IndoTERM-E and its database

3.1 How electricity works in the model

IndoTERM-E is a TERM model: a bottom-up multi-regional CGE in which every region has its own industries, households, investors and government, and goods are traded between regions on delivered prices. The energy extension of February 2026 adds fuel, energy and capital-energy nests in production and CO₂ accounting by fuel and user. Electricity has two layers (Figure 3). Seven generator industries in each region (coal, gas, diesel, hydro, geothermal, solar, wind) make a single commodity, ElecGen; they are substitutes with an elasticity of 10. ElecGen is sourced across regions by a CES choice with elasticity 1.0, and bought by the ElecDist industry, which sells ElecDist (transmission, distribution and retail) to all users.

Regional sourcing (the step our policy touches). For each commodity c, source s (domestic or imported), origin region r and destination d, the model writes the percentage change in the quantity shipped as

$x_{c,s,r,d} - a_{c,s,r,d} = x_{use} + t_{c,s,r,d} - {\overline{t}}_{c,s,d} - \sigma_{c}\left( p_{del} + a_{c,s,r,d} - p_{use} \right)$ (1)

Here x is the quantity, a the iceberg-cost shifter (atrad), t the sourcing-taste shifter (srctwist), σ the sourcing elasticity (SIGMADOMDOM), p_del the delivered price from r to d and p_use the price index of the regional composite (TAB lines 1349–1362). The index is a value-share-weighted average:

$p_{use} = \sum_{r}^{}{S_{r}\left( p_{del} + a \right)}$ (2)

A negative shock to a for Java-origin electricity lowers the delivered price of Java power in destination d and, with elasticity σ, shifts sourcing towards Java. Because electricity is traded by value share and σ is 1.0, the shift is small unless the shock is large (section 6).

3.2 What the database contains

The database is the IndoTERM 2016 IO table (185 sectors) aggregated to 28 commodities, 34 industries and six regions: Sumatera, Java-Bali, Kalimantan, Sulawesi, Nusa Tenggara-Maluku and Papua. As the team recalls, it was calibrated to 2022 GDP; the TABLO file still carries a 2016 exchange rate and “2016 price” labels, so the price year needs confirming. The electricity sector was split from a single sector using PLN's 2022 generation-cost shares by technology (elecost.tab). Generation by technology and region is in the header REEG, which is labelled GWh but is in MWh (it sums to 353 TWh) ^[C14].

Table 4. Electricity facts in the database (computed from aggmod.har)

4. What the model cannot see: issues to resolve before any run

Reading the TAB file and the database against the planning documents shows eight issues. Each is stated as a fact or as a hypothesis to check.

4.1 No surplus and no shortage

The model is demand-driven: output of each industry follows demand, capital is a stock that earns a rental, and there is no engineering capacity limit, reserve margin or dispatch order. Costs per technology are equal across regions, so a Java generator is no cheaper at the margin than an outer-island one. Regional average costs differ only through the technology mix, and these gaps happen to resemble PLN's: the database's average ElecGen price relative to Java is 1.20 (Sumatera), 1.00 (Kalimantan), 1.42 (Sulawesi), 3.13 (NT-Maluku) and 1.11 (Papua), against PLN's 1.21, 1.04 (Kalseltengtimra), 1.55–2.02 (Sulbagsel, Sulbagut), about 3.1 (Ambon, Halmahera) and 2.99 (Jayapura) ^[C13,C18]. Kalbar (2.17) and Papua are the clear mismatches. So a cost gap exists partly, but excess capacity in Java and shortage elsewhere are not represented at all.

4.2 The base-year electricity flows between regions are not physical

The inter-regional trade matrix of TERM is estimated by a gravity-type procedure and RAS scaling, not observed. For electricity it produces flows that cannot exist today: Figure 5 shows 41 TWh of ElecGen going from the outer regions into Java-Bali and 11 TWh going from Java-Bali to the outer regions; Papua sends 9.6 TWh to Java, which is 70% of its generation. The ElecDist matrix has the same problem: 15% of Sumatera's ElecDist use and 43% of Papua's are sourced from Java-Bali ^[C16]. The only physical inter-island link is Java–Bali (inside one region of the database).

Two consequences. A policy that “enables” a link would act on an artificial starting matrix. And because TERM changes sourcing in percentages, a flow that is zero in the base can never become positive: to simulate a new link the base needs small seed flows, or the database must be built for the connected world and the link switched off.

4.3 Electricity sourcing is very inelastic

SIGMADOMDOM for ElecGen and ElecDist is 1.0 ^[C17]. With constant value shares, a cost cut raises the quantity by about the same percentage, so large flows need either a large shock or a higher elasticity. Section 6 works through the arithmetic.

4.4 Captive coal is missing

CO₂ from the MetalProds industry is 0.6 Mt in Sulawesi, 0.04 Mt in NT-Maluku and 14.2 Mt in Java-Bali ^[C15]. The 2016 data predate the nickel smelter boom, and captive plants are in the IO table, if at all, as own-use fuel of the metal industry. The captive 19.3 GW in 2025 ^[C10] therefore has no counterpart in the base. (Hypothesis to check: how the IO table books captive power.)

4.5 Emission factors, units and generation mix need checking

Table 5. Reality check on emissions and generation (to be settled with PLN data)

The RUPTL-implied intensities (1.20 for coal and 0.77 for gas) are the plan's CO₂ rows divided by its energy by fuel (Table 5.143); with the mix row of Table 5.86 instead they would be 1.36 and 0.55, because the two RUPTL tables disagree on the Java-Bali energy mix. They differ from the official grid factors and from the database, and need an explanation from PLN (basis of the emissions rows, gross or net energy, heat rates). Java's hydro, 0.18 TWh in the database against 2.7 GW installed, and Kalimantan's 14.4 TWh of hydro in the database look inconsistent with the plan; we suspect the regional allocation of REEG and ask the team who built it.

4.6 Summary of the issues

Table 6. Issues and what to do about them

5. What the literature says

Two Undermind deep searches (152 papers on grid interconnection and Indonesian power planning; 170 papers on CGE models with electricity) and targeted reading of the papers you collected give the picture below. For the points cited we put targeted questions to the full text of Langer et al., Sarjiya et al. (2022), Utomo et al. and Nurliani and Noveria, and read the method sections of Horridge et al. and Diniz et al. The others are known from abstracts only ^[C21].

5.1 Studies of inter-island links and Indonesian power planning

Table 7. Closest studies on links and Indonesian power systems

Further reading: Indonesian studies and the direction of links in Appendix D2, group 1a and 1b.

5.2 Why models move power into Java, and what we can learn

In the decarbonisation studies the flow into Java comes from least-cost design: Java has high demand and little land for onshore renewables, while hydro, geothermal and solar are cheaper elsewhere (Langer et al.). Those scenarios retire fossil capacity to zero, so Java's coal surplus no longer exists at the end of the period; the RUPTL's ISJ has the same logic. This does not contradict the existence of a surplus today. It means the direction of value depends on the period: before large renewable build-out, surplus Java capacity can supply outward; later, outer-island renewables supply Java. From these studies we can take: the cost and loss parameters of HVDC links (US$171,000 per MW for the converter pair plus US$293 per MW-km for the cable, 1.7% annual operating cost, 96.5% efficiency per 1,000 km) ^[C20]; the need to treat captive and grid supply separately; and the warning that interconnection alone is not clean.

5.3 When does a link lower emissions?

Two studies of coal-heavy systems find that transmission can raise emissions. In China, expanding inter-regional transmission raised power-sector CO₂ by about 2% in 2030 because it let cheap, low-efficiency coal supply more regions (Li et al. 2016). In Europe, transmission raises emissions when renewable output is low, because it moves coal generation (Abrell and Rausch 2016) ^[C22]. Per MWh delivered, the effect of a link is

$\Delta E = {EF}_{Java}(1 + \lambda) - {EF}_{displaced}$ (3)

where λ is the line loss (6.2% for 1,680 km of HVDC at 96.5% per 1,000 km) and the emission factors are those of the marginal plants in Java and of the plants displaced outside Java. Figure 6 maps the sign. The team note reaches the same formula (its section 4.4).

Three lessons for the design. A link replacing outer-island coal or diesel and fed by Java coal is roughly neutral (0.80 × 1.062 against 0.87). It lowers emissions where the displaced grid is dirtier than Java's, such as the Kalimantan grids (1.12–1.67 in 2019) or Lombok (1.27). It raises emissions where it displaces gas or renewables, or if Java's marginal unit is coal while a gas unit would have been displaced. Whether the marginal Java unit is coal or gas is exactly the question the low use of Java's gas plants raises.

Further reading: evidence on when links raise or cut emissions in Appendix D2, group 1c.

5.4 CGE precedents

• Horridge, Yusuf and Ginting (2016, ADB Papers on Indonesia No. 17): the closest template. IndoTERM, medium/long-run closure (regional employment fixed through wages, rates of return fixed, trade balance to GDP fixed, exchange rate as numeraire). Connectivity is simulated as a 5% cut in road and sea transport margins plus a 1% cut in iceberg cost. The full portfolio gives benefits of about 5% of GDP (p. 3); in the first component most of the national GDP gain (0.96 of 1.09%) comes from the 1% iceberg cut (p. 9). The authors simulate only the benefits, “just half of a full cost-benefit analysis” (p. 6) ^[C26].

• Diniz, Ferreira-Filho and Horridge (2018, TERM-BR10, working paper marked “not to be cited”): a recursive-dynamic TERM with seven generation industries and a transmission-and-distribution industry that collects local generation and passes the surplus to other regions by inter-regional trade, the same structure as ours. National plan capacity by source, region and year is turned into investment shocks; a baseline plan is compared with alternative plans (pp. 7–10) ^[C26]. We use it for method and should ask the authors before citing.

• Abrell and Rausch (2016): a general-equilibrium model with explicit transmission constraints, an example of combining electricity dispatch with economy-wide effects.

• GTAP-Power and technology-bundle papers (Peters 2016; Chepeliev 2023; Cai and Arora 2015; Cui et al. 2020): transmission and distribution as a separate activity and the effect of nesting choices on results; Cui et al. warn that results depend on the substitution elasticities between power sources.

• Indonesian CGE with energy: Hasudungan and Sabaruddin (2018, feed-in tariffs), Hartono, Komarulzaman et al. (2020, energy subsidies, recursive dynamic), Yusuf et al. (2023, capital city and carbon, inter-regional). None models inter-island electricity flows.

Further reading: CGE and economy-wide models of interconnection and electricity investment in Appendix D2, group 2.

5.5 The gap

No study found measures whether moving Java-Bali surplus power to outer-island grids avoids new utility or captive coal, nor its CO₂ and economy-wide effect ^[C21]. The team note also finds no study of the emission effect of the ISJ. This is the contribution available to us, with the caveat of section 2.6.

5.6 Costs and wider benefits of a link

A plant and a link are not the same purchase. A plant makes energy and needs fuel; a link moves energy and needs a supplier at the other end, here Java's surplus. We therefore compare them twice: capital cost per MW of capacity, and cost per MWh delivered. The case is the eastern corridor (Java–Bali–Lombok), where RUPTL 2025–2034 plans 200 MW of new gas engines in Lombok for 2027–2030 (Lombok 3 and 4) and lists a 150 kV Bali–Lombok link of about 74 km (overhead line and sea cable) as still under study (p. V-322 and Lampiran C11).

Three readings. A short sea link costs about as much per MW as new gas engines (0.37–0.81 against 0.80 million USD per MW) and less than a new coal plant (1.36–1.60). Literature costs for long HVDC links are low, but Indonesian project estimates are not: the ISJ at 2.5 and Sumba–Java at about 4.4 million USD per MW cost more per MW than a coal plant. Long links therefore have to earn their keep through the energy they bring and the benefits below, not through lower capital cost.

On these assumptions the link delivers power to Lombok at about 74–84 USD per MWh against 124–156 for new gas engines. Four cautions. First, a single link can fail, so Lombok still needs local reserve and not all 200 MW of engines is avoided. Second, the Java side is priced at its average cost; the marginal cost, the free capacity on the Java–Bali 500 kV link (COD 2030) and the end of Java's surplus around 2028–2030 decide the real figure. Third, RUPTL flags the deep sea trench between Bali and Lombok, which raises cost and risk. Fourth, the cost sources use different price years. The result is a screening estimate, to be replaced by PLN's own feasibility numbers.

Benefits beyond money saved. Table 5.6 lists what a link brings that a cost comparison misses, with the evidence behind each item. Most are conditional on what the link displaces at the receiving end and what it calls on at the sending end.

Sources for this section are grouped in Appendix D2, groups 1d and 1e.

6. Designing the static simulation

6.1 What a static run is, and when it makes sense

A static (comparative-static) run solves the model once for a chosen closure and shows the new equilibrium compared with the base, with no time path. It is useful for three things: checking that the levers do what we intend; decomposing channels cleanly; and giving a target-year snapshot once demand growth has been imposed. It cannot by itself show avoided new coal, because in a long-run closure capital adjusts at once and there is no baseline growth path. We therefore propose it as a stage before the dynamic runs, not as the headline.

6.2 Closures

Table 8. Candidate closures

Common to all: exchange rate as numeraire, balance of trade to GDP fixed, national household consumption following income. The dynamic closures in the package must be reversed for a static run (capital accumulation, dynamic investment rule and the demonstration shocks removed).

6.3 How the Java surplus and the outer-island shortage enter

Because the model ignores them (section 4.1), the situation is created by the scenario:

• Outer-island growth counterfactual (the shortage). Add electricity demand growth outside Java, for example nickel and metal output in Sulawesi and NT-Maluku and household growth, from RUPTL peak-load projections. In the no-link world, a long-run closure meets it with more outer-island generator capital: this is the 'new coal' reading. The link scenario has the same growth plus the connection; the effect is the difference between the two runs.

• Idle Java capacity (the surplus). Fix Java generator capital so that extra output does not carry a new capital cost, or raise capital-augmenting technical change (acap) in Java generators by the size of the surplus. With capital fixed, extra output is made by substituting labour and fuel (elasticity 1.26), so marginal cost rises a little; this is checked by sensitivity.

• Regional cost gap (optional). Table 2 gives PLN's cost by system. The database already reproduces part of the gap through the technology mix; a BPP-based adjustment would remove the mismatches (Kalbar, Papua). This changes the database and needs the team's agreement.

6.4 Options for the link itself

Table 9. Options A–C (the team decided to start with A as a mechanics test, then B and C)

6.5 A worked example from the database

Java supplies 10.8% of Sumatera's ElecGen by value, equal to 7.2 TWh at Java's average price of Rp 705/kWh ^[C16]. In a partial-equilibrium reading of equations (1) and (2), with other prices and total demand held fixed, a cut of a percent in the delivered cost of Java power raises the delivered quantity from Java by

$\frac{\Delta Q_{J}}{Q_{J}} \approx \sigma \cdot a \cdot \left( 1 - S_{J} \right)$ (4)

With S_J=0.108: a 10% cut gives +9% with σ=1 (0.8 TWh), +22% with σ=2.5 and +45% with σ=5 (3.2 TWh) (Figure 7). A link of 2.6 GW at 70% load, the planned ISJ transfer, carries about 16 TWh a year, or +223% of the base flow. With σ=1 no plausible cost cut achieves that; with σ=5 it needs a cut of about 50%. So the shock and the elasticity must be calibrated to the physical link, and the base flow must be re-based (section 4.2).

Calibration chain: link capacity (GW) × load factor × 8.76 = TWh; minus line losses; divided by the base flow = percentage change; solved for the shock given σ. Using the ISJ capacity, 2.6 GW, gives 15.9 TWh at 70% load; the 3.2–4.9 GW surplus of 2025 (Figure 1) gives 20–30 TWh at the same load.

6.6 Static scenario sequence

• S0 zero-shock run reproduces the base data (assertions, updated data equal to input).

• S1 mechanics test of option A: atrad −10% Java→Sumatera in the short- and long-run closures; check flows, emissions bookkeeping and the energy balance.

• S2 calibrated A for the ISJ-sized link, then option B with the outer-island growth counterfactual, then A+B.

• S3 extend to Kalimantan, Sulawesi and NT-Maluku/Papua; reverse direction (outer→Java) as a mirror case.

• S4 sensitivities: σ for ElecGen (1, 2.5, 5, 10); ElecDist cross-region trade removed; emission factors replaced by official ones; share of captive demand connected; carbon price on and off; closure.

7. Brainstorm: the dynamic IndoTERM-E simulation

7.1 Why the dynamic model suits the question

The question is about avoided new coal and the timing of a link against a shrinking surplus. The recursive-dynamic model has a capital accumulation mechanism (capital grows by investment less depreciation, with a one-year lag), an investment rule linking investment to expected returns, and a real-wage adjustment, so a baseline path with outer-island growth and a policy path with the link can be compared year by year to 2050. Cumulative emissions, the commissioning year and the effect of coal built before the link are all visible, which a static run cannot show.

7.2 Baseline

• Start from the existing GMMI dynamic set-up (the package files COMMON.CLS, BASE.BSH, POL2.CLS and the ALLINV files; 28 periods, 2023–2050). The earlier GMMI baseline grows alnd, labour productivity, households, trend employment and export quantity uniformly.

• Add the growth that creates the outer-island shortage: metal and nickel output in Sulawesi and NT-Maluku, households and services, from RUPTL peak-load growth by system (12.4 GW across five outer systems against 14.8 GW in Java-Bali, 2025–2034).

• Impose capacity additions by technology, region and year for the electricity industries from the RUPTL scenario chosen (RE Base, ARED or BAU PLTU), as POL2.CLS already does for renewable investment (xinvitot exogenous for selected generators) and as Diniz et al. did. Convert MW to rupiah using the model's own capital stock per MW; check it against the capital cost per kW (the database holds Rp 241 trillion of coal-power capital, to be set against installed coal capacity).

• Choose the baseline plan: RE Base is the plan PLN judges most likely; ARED is the accelerated renewable plan; the BAU PLTU case is a reference.

7.3 Captive coal

Because the 2016 base has almost none, the baseline needs a deliberate treatment. Three options:

Table 10. Treatments of captive coal (decision D5)

7.4 Policy shocks

• The link: a ramp of atrad and srctwist over the commissioning years (RUPTL: ISJ 2031, JBC cable 2030; a news report of 1 October 2026 quotes a 2032–2035 target for inter-island links, secondary), scaled by GW commissioned and converted from GW to TWh by load factor.

• Avoided coal: lower the exogenous investment path of outer-island coal units that the link makes unnecessary (utility units first; captive units only under the share-connected assumption); the units with signed PPAs (Jambi-1, Jambi-2) are marked as costly to cancel.

• Java generators: utilisation rises when Java supplies outward. Report the ratio of output to capital for Java generators and bound it by a check, or keep Java capital fixed in the closure.

• The cost of the link: add the investment (about US$4.28 bn for the Sumatera–Java route per ESDM; converter and cable cost per Langer et al.) as construction demand and as capital in ElecDist, financed under the chosen closure. Without it only benefits are shown.

• Direction: run a second scenario with the later phase in which the link evacuates outer-island renewables to Java.

7.5 Closures, macro and other considerations

• Investment and savings: national investment follows the baseline (invslack exogenous, as in POL2.CLS) or follows GDP (COMMON.CLS); the choice decides whether link investment crowds out other investment.

• External balance: balance-of-trade share of GDP fixed (houslack) or exchange-rate adjustment (TERM-Brazil let foreign savings finance electricity investment, causing appreciation).

• Labour: national real-wage adjustment mechanism (delfwage) with trend employment; regional migration by wage differences.

• Rebound and leakage: cheaper outer-island power may expand energy-intensive industry there; report emissions by region and by user, not only by the power sector. The team note's ‘Sumatera paradox’ (Sumatera exports to Java while adding mine-mouth coal) shows why the combined emissions of the connected regions are the right measure.

• Java coal retirement: the surplus depends on Java's coal fleet; run the link under two paths (fleet retained; early retirement under the JETP plan; press reports say the early retirement of Cirebon-1 was dropped, a secondary source).

• Carbon price: with and without delco2taxrate, since Atmo et al. find that carbon pricing changes the sign for ASEAN trade.

• Baseline calibration: compare baseline CO₂ with the official energy GHG inventory and RUPTL emissions (309 to 374 Mt for PLN plants 2025–2034 in the team note) before using it.

• Technical: RunDynam in C:\gmmi\regdyn with Base, Rerun and Policy names; results combined with cmbhar; start with a zero-shock policy file to confirm policy equals baseline.

7.6 Outputs to report

CO₂ by region and by user, annual and cumulative 2025–2050, for baseline and policy; generation by technology and region; avoided outer-island coal and diesel capital; GDP, consumption and real wages by region; electricity price by region; and the regime map of section 5.3 filled in with model results. Where the model cannot produce a number (cost of supply to PLN, subsidy), say so.

8. Scenario matrix and sensitivities

Table 11. Scenario dimensions

The full factorial is too large. We propose a core set (link capacity and timing, σ, displaced supply, Java marginal supply) and one-at-a-time variation of the others around it.

9. Data, calibration and verification plan

• Evidence ledger from the RUPTL (Kepmen ESDM 188.K/TL.03/MEM.L/2025), PLN statistics, RUKN and official emission factors: reserve margin and peak by system, capacity and generation by region and technology, planned additions, links with GW and direction, BPP by system. Every figure with page and link.

• Open items already known: capacity of ISJ per direction and whether bidirectional; heat rates by plant; actual reserve margins in 2025–2026; the two ESDM news items (oversupply and the March 2023 Java→Sumatera proposal), which we could read only through search summaries; take-or-pay terms; the basis of the RUPTL emission rows.

• Model checks: zero-shock reproduction; accounting identities; electricity balance (change in generation equals TWh moved); CO₂ bookkeeping (change in national emissions equals the sum over regions and users, with the power part split by fuel); comparison with Langer et al. (cost effects) and CREA/GEM (captive scale).

• Reporting rule: every number in a deliverable has a source, page or table, and link; results from the model are reported only after the checks pass.

10. Considerations at a glance, risks and work plan

Table 12. Considerations at a glance

10.1 Risks and work plan

Table 13. Main risks

Table 14. Proposed work plan (division of work to be agreed)

Appendix A. Key relations of the model used in this paper

Equations (1) and (2) are in section 3.1 and equation (4) in section 6.5. Two further relations matter for the dynamic run. Capital accumulates with a one-year lag:

$K_{i,t + 1} = \left( 1 - \delta_{i} \right)K_{i,t} + I_{i,t}$ (5)

where δ is the depreciation rate (5% for every industry in the database) ^[C17]. Planned investment per unit of capital rises with the expected rate of return relative to the normal rate (normal gross return 13.8%, trend investment/capital ratio 9%, maximum ratio five times the trend, investment elasticity 5, partial adjustment of expected returns 0.5; the last four are identical for coal and solar in every region) (TAB excerpt 59, headers TARG, TFRO, QRAT, ALFA, RADJ). In the policy run the investment of selected electricity industries is set exogenously from the plan instead.

Appendix B. Database tables

Table B1. Generation by technology and region (header REEG, read as MWh) ^[C14]

Table B2. CO₂ by generator type and region, and the metal industry (header QGAS) ^[C15]

Table B3. ElecGen sourcing matrix in TWh (header TRAD, Rp billion converted at the origin's average price) ^[C16]

Appendix C. Numbers register

Each tag such as [C3] used in the text is listed with its source and where to find it. 'Computed' means calculated by us from the source shown; the scripts are tools/db_electricity_tables.py and tools/make_figures.py. Files named under 'team Dropbox' are in the shared GMMI/referensi folder. Items marked secondary were not read in the primary document.

Appendix D. References

Where a paper has an open PDF it is in your 'add ref' folder or in Undermind. DOIs as returned by Undermind; author lists for Langer et al. and Gulagi et al. are shortened in the database and will be completed from the DOI pages.

Abrell, J. and Rausch, S. (2016). Cross-country electricity trade, renewable energy and European transmission infrastructure policy. Journal of Environmental Economics and Management 79, 87–113. doi:10.1016/J.JEEM.2016.04.001 · OpenAlex W2300751791, cited by 142

Amiruddin, A., Dargaville, R. and Gawler, R. (2024). Optimal integration of renewable energy, energy storage, and Indonesia's super grid. Energies 17(20), 5061. doi:10.3390/en17205061 · OpenAlex W4403398778, cited by 11

Atmo, G., Otsuki, T. and Nurcahyanto, E. (2022). Modelling low carbon electricity generation of an integrated ASEAN power grid. IOP Conf. Ser.: Earth Environ. Sci. 997, 012011. doi:10.1088/1755-1315/997/1/012011 · OpenAlex W4214940468, cited by 8

Boys, K.A., Willis, D.B. and Rum, I.A. (2012). The impact of trade costs in Indonesian agri-food sectors: an interregional CGE analysis. doi:10.22004/AG.ECON.124797 · OpenAlex W1576934883, cited by 2

Cai, Y. and Arora, V. (2015). Disaggregating electricity generation technologies in CGE models: a revised technology bundle approach with an application to the U.S. Clean Power Plan. Applied Energy 154, 543–555. doi:10.1016/J.APENERGY.2015.05.041 · OpenAlex W1760483588, cited by 55

Chepeliev, M. (2023). GTAP-Power data base: version 11. Journal of Global Economic Analysis 8(2). doi:10.21642/jgea.080203af · OpenAlex W4389720055, cited by 14

CREA and Global Energy Monitor (2026). Out of sight, out of control: 31 GW of captive coal is jeopardizing Indonesia's economic and emissions goals. PDF

Cui, Q., Liu, Y., Ali, T., Gao, J. and Chen, H. (2020). Economic and climate impacts of reducing China's renewable electricity curtailment: a comparison between CGE models with alternative nesting structures of electricity. Energy Economics. doi:10.1016/j.eneco.2020.104892 · OpenAlex W3047062578, cited by 93

Diniz, T., Ferreira-Filho, J.B.S. and Horridge, M. (2018). Economic and regional impacts of investments in electricity generation in Brazil. Working paper (marked 'not to be cited'). Semantic Scholar

Gulagi, A., Oyewo, A.S., Bogdanov, D., et al. and Breyer, C. (2025). Accelerating the transition from coal to renewables in Indonesia to achieve a net-zero energy system. IET Renewable Power Generation. doi:10.1049/rpg2.13188 · OpenAlex W4407097707, cited by 12

Hartono, D., Komarulzaman, A., Irawan, T. and Nugroho, A. (2020). Phasing out energy subsidies to improve energy mix: a dead end. Energies 13(9), 2281. doi:10.3390/en13092281 · OpenAlex W3023074185, cited by 15

Hasudungan, H. and Sabaruddin, S.S. (2018). Financing renewable energy in Indonesia: a CGE analysis of feed-in tariff schemes. Bulletin of Indonesian Economic Studies 54(2), 233–264. doi:10.1080/00074918.2018.1450961 · OpenAlex W2793657753, cited by 15

Horridge, M., Yusuf, A.A., Ginting, E. and Aji, P. (2016). Improving Indonesia's domestic connectivity: an inter-regional CGE analysis. ADB Papers on Indonesia No. 17. Asian Development Bank.

Langer, J., et al. and Blok, K. (2024). The role of inter-island transmission in full decarbonisation scenarios for Indonesia's power sector. Environmental Research: Energy. doi:10.1088/2753-3751/ad53cb · OpenAlex W4399331911, cited by 10

Lanz, B. and Rausch, S. (2011). General equilibrium, electricity generation technologies and the cost of carbon abatement. MIT Joint Program Report 194. Semantic Scholar

Li, T., Liu, P. and Li, Z. (2019). Quantifying cross-provincial power transmission barriers in China, based on a computable general equilibrium approach. Energy Procedia. doi:10.1016/J.EGYPRO.2019.01.136 · OpenAlex W2921452830, cited by 2

Li, Y., Lukszo, Z. and Weijnen, M. (2016). The impact of inter-regional transmission grid expansion on China's power sector decarbonization. Applied Energy 183, 853–873. doi:10.1016/J.APENERGY.2016.09.006 · OpenAlex W2525511425, cited by 94

Nurliani, A. and Noveria, A. (2025). Techno-economic analysis of the transition from off-grid to on-grid electricity in the mining industry. E-BISMA.

Ordonez, J.A., Fritz, M. and Eckstein, J. (2022). Coal vs. renewables: least-cost optimization of the Indonesian power sector. Energy for Sustainable Development. doi:10.1016/j.esd.2022.04.017 · OpenAlex W4229335981, cited by 51

PT PLN (Persero) (2025). Rencana Usaha Penyediaan Tenaga Listrik 2025–2034, ratified by Kepmen ESDM No. 188.K/TL.03/MEM.L/2025.

Peters, J.C. (2016). GTAP-E-Power: an electricity-detailed economy-wide model. Journal of Global Economic Analysis 1(2), 156–187. doi:10.21642/JGEA.010204AF; and The GTAP-Power data base, 1(1), 209–250. doi:10.21642/JGEA.010104AF · OpenAlex W2565090201, cited by 47; OpenAlex W2410475769, cited by 110

Sarjiya, Budi, R., et al. (2022). Assessing potential scenarios for achieving new and renewable energy targets in Java-Bali power system, Indonesia. International Journal of Energy Economics and Policy. doi:10.32479/ijeep.12852 · OpenAlex W4221002146, cited by 11

Sarjiya, Putranto, L.M., Tumiran, Budi, R., Novitasari, D. and Deendarlianto (2023). Generation expansion planning with a renewable energy target and interconnection option: a case study of the Sulawesi region, Indonesia. Renewable and Sustainable Energy Reviews 183, 113489. doi:10.1016/j.rser.2023.113489 · OpenAlex W4382877043, cited by 25

Silalahi, D.F., Blakers, A. and Cheng, C. (2024). 100% renewable electricity in Indonesia. Energies 17(1), 3. doi:10.3390/en17010003 · OpenAlex W3110131985, cited by 22

Utomo, D.P., Subrata, M.E. and Dhelika, R. (2025). Decarbonization roadmap and energy system scenarios for Southern Sulawesi system: response to demand growth from smelters sector. Journal of Asian Energy Studies. doi:10.24112/jaes.090009 · OpenAlex W4411720488, cited by 1

Wing, I.S. (2008). The synthesis of bottom-up and top-down approaches to climate policy modeling: electric power technology detail in a social accounting framework. Energy Economics 30, 547–573. doi:10.1016/J.ENECO.2006.06.004 · OpenAlex W2026018101, cited by 152

Yuwono, B., Kranzl, L., Haas, R., Dewi, R.G., Siagian, U.W., Kraxner, F. and Yowargana, P. (2025). Incorporating grid development in capacity expansion optimisation: a case study for Indonesia. Applied Energy. doi:10.1016/j.apenergy.2024.124837 · OpenAlex W4404433513, cited by 9

Yusuf, A.A., Roos, E., Horridge, J.M. and Hartono, D. (2023). Indonesian capital city relocation and regional economy's transition toward less carbon-intensive economy: an inter-regional CGE analysis. Japan and the World Economy 68, 101212. doi:10.1016/j.japwor.2023.101212 · OpenAlex W4386913471, cited by 22

GMMI (2025). GMMI Round 2 Scenario Specification, 28 November 2025.

Appendix D2. Literature support from OpenAlex

Searched in OpenAlex on 16 October 2026; every new entry was checked against its abstract. Group 1 covers the relevance of, and rationale for, connecting grids. Group 2 covers CGE and other economy-wide models with the same research aim. new marks works not yet in Appendix D; in Appendix D marks works already cited. Citation counts are OpenAlex counts on that date.

Group 1. Relevance and rationale of grid connection

1a. Indonesia: surplus, shortage, captive coal and the direction of links

• in Appendix D Jannis Langer et al. (2024). The role of inter-island transmission in full decarbonisation scenarios for Indonesia’s power sector. Environmental Research Energy. doi:10.1088/2753-3751/ad53cb · OpenAlex W4399331911, cited by 10.
  Least-cost decarbonisation with about 43 GW of inter-island lines, mostly into Java; without links Java can still be supplied, at about 15% higher cost. See section.
• in Appendix D Sarjiya Sarjiya et al. (2022). Assessing Potential Scenarios for Achieving New and Renewable Energy Targets in Java-Bali Power System, Indonesia. International Journal of Energy Economics and Policy. doi:10.32479/ijeep.12852 · OpenAlex W4221002146, cited by 11.
  A Nusa Tenggara to Java-Bali HVDC link appears in one of its generation-expansion scenarios. See section.
• in Appendix D Sarjiya Sarjiya et al. (2023). Generation expansion planning with a renewable energy target and interconnection option: A case study of the Sulawesi region, Indonesia. Renewable and Sustainable Energy Reviews. doi:10.1016/j.rser.2023.113489 · OpenAlex W4382877043, cited by 25.
  Generation expansion for Sulawesi with an interconnection option. See section.
• in Appendix D Bintang Yuwono et al. (2024). Incorporating grid development in capacity expansion optimisation - a case study for Indonesia. Applied Energy. doi:10.1016/j.apenergy.2024.124837 · OpenAlex W4404433513, cited by 9.
  Adds grid development to capacity-expansion optimisation for Indonesia. See section.
• in Appendix D Ahmad Amiruddin, Roger Dargaville and Ross Gawler (2024). Optimal Integration of Renewable Energy, Energy Storage, and Indonesia’s Super Grid. Energies. doi:10.3390/en17205061 · OpenAlex W4403398778, cited by 11.
  Optimises renewables, storage and an Indonesian super grid together. See section.
• in Appendix D Ashish Gulagi et al. (2025). Accelerating the transition from coal to renewables in Indonesia to achieve a net‐zero energy system. IET Renewable Power Generation. doi:10.1049/rpg2.13188 · OpenAlex W4407097707, cited by 12.
  Coal-to-renewables pathway for a net-zero Indonesian energy system. See section.
• in Appendix D David Firnando Silalahi, Andrew W Blakers and Cheng Cheng (2023). 100% Renewable Electricity in Indonesia. Energies. doi:10.3390/en17010003 · OpenAlex W3110131985, cited by 22.
  A 100% renewable electricity system for Indonesia. See section.
• in Appendix D Jose Antonio Ordonez, Marek Fritz and Johannes Eckstein (2022). Coal vs. renewables: Least-cost optimization of the Indonesian power sector. Energy Sustainable Development/Energy for sustainable development. doi:10.1016/j.esd.2022.04.017 · OpenAlex W4229335981, cited by 51.
  Least-cost optimisation of the Indonesian power sector, coal against renewables. See section.
• in Appendix D Gigih Udi Atmo, Takashi Ôtsuki and Eri Nurcahyanto (2022). Modelling low carbon electricity generation of an integrated ASEAN power grid. IOP Conference Series Earth and Environmental Science. doi:10.1088/1755-1315/997/1/012011 · OpenAlex W4214940468, cited by 8.
  Low-carbon generation with an integrated ASEAN grid. See section.
• new Appin Purisky Redaputri (2023). The Condition of Excess Electricity Supply in Indonesia. Journal of Earth Energy Engineering. doi:10.25299/jeee.2023.12856 · OpenAlex W4388233421, cited by 3.
  Describes the move from shortage (2015) to oversupply after the 35,000 MW programme met stagnant demand; supports the Java surplus premise. See section.
• new Paul J. Burke and Sandra Kurniawati (2018). Electricity subsidy reform in Indonesia: Demand-side effects on electricity use. Energy Policy. doi:10.1016/j.enpol.2018.02.018 · OpenAlex W2793690086, cited by 95.
  Electricity subsidy reform reduced electricity use, one reason demand fell short of the growth the 35 GW programme was sized for. See section.
• new Muhammad Zulfizal Arnaz (2019). THE AGGREGATE AND RESIDENTIAL ELECTRICITY DEMAND IN INDONESIA. Jurnal Ekonomi dan Pembangunan. doi:10.14203/jep.26.1.2018.1-20 · OpenAlex W2913846647, cited by 4.
  Income and customer numbers drive electricity demand elastically; demand in Java-Bali responds less to price than outside Java. See section.
• new Angela Tritto (2021). China's Belt and Road Initiative: from perceptions to realities in Indonesia's coal power sector. Energy Strategy Reviews. doi:10.1016/j.esr.2021.100624 · OpenAlex W3131356760, cited by 54.
  Documents a wave of 'instrumental' coal plants serving industrial parks and owned by non-energy firms, mostly subcritical: the captive coal the model lacks. See section.
• new Bowen Gu (2024). Black gold and green BRI–A grounded analysis of Chinese investment in coal-fired power plants in Indonesia. The Extractive Industries and Society. doi:10.1016/j.exis.2024.101411 · OpenAlex W4391671811, cited by 12.
  Maps 25 conflicts over coal plants in Indonesia, including plants tied to nickel processing. See section.
• new Wenjing Wei et al. (2020). Energy Consumption and Greenhouse Gas Emissions of Nickel Products. Energies. doi:10.3390/en13215664 · OpenAlex W3095342921, cited by 108.
  GHG intensity of nickel products depends strongly on the electricity source (nickel pig iron about 69 t CO2e per t of nickel): why captive coal matters for smelter regions. See section.
• new Thang Nam and Paul J. Burke (2024). Phasing out coal power in two major Southeast Asian thermal coal economies: Indonesia and Vietnam. Energy Sustainable Development/Energy for sustainable development. doi:10.1016/j.esd.2024.101451 · OpenAlex W4396530863, cited by 29.
  Finds that halting new coal plants is more practical than early closures in Indonesia: the margin a Java link would act on. See section.

1b. Why interconnect: evidence from other systems

• new Werner Antweiler (2016). Cross-border trade in electricity. Journal of International Economics. doi:10.1016/j.jinteco.2016.03.007 · OpenAlex W2188760714, cited by 89.
  Two-way electricity trade as insurance against correlated demand shocks ('reciprocal load smoothing'); an economic rationale for a supergrid. See section.
• new Fridrik M. Baldursson et al. (2018). Cross-Border Exchange and Sharing of Generation Reserve Capacity. The Energy Journal. doi:10.5547/01956574.39.4.fbal · OpenAlex W2764154104, cited by 28.
  Sharing generation reserves across borders cuts procurement costs (about EUR 500 million a year in six EU countries): the reserve-margin argument for linking a surplus system. See section.
• new Åsa Grytli Tveten, Jon Gustav Kirkerud and Torjus Folsland Bolkesjø (2016). Integrating variable renewables: the benefits of interconnecting thermal and hydropower regions. International Journal of Energy Sector Management. doi:10.1108/ijesm-08-2014-0006 · OpenAlex W2510299658, cited by 23.
  Linking thermal- and hydro-dominated regions cuts curtailment of variable renewables: the Sumatra hydro to Java thermal case. See section.
• new Xu Deng et al. (2022). Assessing the integration effect of inter-regional transmission on variable power generation under renewable energy consumption policy in China. Energy Policy. doi:10.1016/j.enpol.2022.113219 · OpenAlex W4293431226, cited by 68.
  Assesses how inter-regional transmission helps integrate variable renewables under China's renewable consumption policy. See section.
• new Michalis Karystianos et al. (2021). Planning of Aegean Archipelago Interconnections to the Continental Power System of Greece. Energies. doi:10.3390/en14133818 · OpenAlex W3175528065, cited by 10.
  Grid operator planning for linking Aegean islands to the Greek mainland, to retire costly oil units and raise renewables: an archipelago analogue for Nusa Tenggara and Maluku. See section.
• new Javier Bustos-Salvagno and Fernando Fuentes H. (2017). Electricity Interconnection in Chile: Prices versus Costs. Energies. doi:10.3390/en10091438 · OpenAlex W2754812535, cited by 11.
  Joining Chile's two largest systems lowers contract prices through less cost variability and less market concentration. See section.
• new Huaibo Yang et al. (2022). Has the Inter-Regional Power Transmission Promoted Economic Development? A Quantitative Assessment in China. Sustainability. doi:10.3390/su142013402 · OpenAlex W4306769724, cited by 7.
  County panel (2000-2019): China's UHV inter-regional transmission raised growth along the lines, especially in power-exporting provinces. See section.
• new Han Jiang et al. (2019). Study of future power interconnection scheme in ASEAN. Global Energy Interconnection. doi:10.1016/j.gloei.2020.01.009 · OpenAlex W3004225962, cited by 19.
  ASEAN interconnection scheme to match remote clean resources with load centres. See section.
• new Yu Wen Huang, Noah Kittner and Daniel M. Kammen (2019). ASEAN grid flexibility: Preparedness for grid integration of renewable energy. Energy Policy. doi:10.1016/j.enpol.2019.01.025 · OpenAlex W2914900786, cited by 105.
  Readiness of ASEAN grids to integrate variable renewables. See section.

1c. When does a link cut emissions?

• in Appendix D Ying Li, Zofia Lukszo and Margot Weijnen (2016). The impact of inter-regional transmission grid expansion on China’s power sector decarbonization. Applied Energy. doi:10.1016/j.apenergy.2016.09.006 · OpenAlex W2525511425, cited by 94.
  In China, inter-regional transmission expansion raised power-sector CO2 by about 2% in 2030 because it let cheap coal supply more regions. See section.
• in Appendix D Jan Abrell and Sebastian Rausch (2016). Cross-Country Electricity Trade, Renewable Energy and European Transmission Infrastructure Policy. Journal of Environmental Economics and Management. doi:10.1016/j.jeem.2016.04.001 · OpenAlex W2300751791, cited by 142.
  In Europe, transmission raises emissions when renewable output is low because it moves coal generation. See section.
• new Joshua Graff Zivin, Matthew J. Kotchen and Erin T. Mansur (2014). Spatial and temporal heterogeneity of marginal emissions: Implications for electric cars and other electricity-shifting policies. Journal of Economic Behavior & Organization. doi:10.1016/j.jebo.2014.03.010 · OpenAlex W3123846525, cited by 384.
  Marginal emissions differ widely across regions and hours; average factors mislead. Supports using marginal plants in equation (3). See section.
• new J.H. Nelson et al. (2012). High-resolution modeling of the western North American power system demonstrates low-cost and low-carbon futures. Energy Policy. doi:10.1016/j.enpol.2012.01.031 · OpenAlex W2077020260, cited by 187.
  High-resolution model of western North America: with a carbon price, transmission expansion supports a low-cost, low-carbon system. See section.

1d. Costs: building new plants against connecting grids

• in Appendix D Jannis Langer et al. (2024). The role of inter-island transmission in full decarbonisation scenarios for Indonesia’s power sector. Environmental Research Energy. doi:10.1088/2753-3751/ad53cb · OpenAlex W4399331911, cited by 10.
  Its appendix gives the HVDC cost basis used in section 5.6: cable 171-1,261 USD per MW per km (median 293) and converter pairs of about 0.15-0.21 million USD per MW, plus generation costs for Indonesia. See section.
• new Abdulrahman Alassi et al. (2019). HVDC Transmission: Technology Review, Market Trends and Future Outlook. Renewable and Sustainable Energy Reviews. doi:10.1016/j.rser.2019.04.062 · OpenAlex W2949851680, cited by 476.
  Review of HVDC technology, market trends and costs, including when HVDC beats AC over distance. See section.
• new Don Hur (2012). Economic Considerations Underlying the Adoption of HVDC and HVAC for the Connection of an Offshore Wind Farm in Korea. Journal of Electrical Engineering and Technology. doi:10.5370/jeet.2012.7.2.157 · OpenAlex W2110268676, cited by 28.
  Discounted-cash-flow comparison of HVDC and HVAC for an 80 km, 2 GW sea link: the method behind choosing AC for short straits and DC for long routes. See section.
• new Alexandra Lüth et al. (2023). How to connect energy islands: Trade-offs between hydrogen and electricity infrastructure. Applied Energy. doi:10.1016/j.apenergy.2023.121045 · OpenAlex W4366609551, cited by 55.
  Joint capacity-expansion and dispatch model: power cables win for sites close to shore, other carriers for far-away islands. Distance decides whether moving energy beats producing it locally. See section.
• new Isaac‐Camilo Gonzalez‐Romero, Sonja Wogrin and Tomás Gómez San Román (2019). Review on generation and transmission expansion co‐planning models under a market environment. IET Generation Transmission & Distribution. doi:10.1049/iet-gtd.2019.0123 · OpenAlex W2955492948, cited by 58.
  Review of models that plan generation and transmission together, the framework in which 'new plant or new link' is a single decision. See section.
• new Eleni Zafeiratou and Catalina Spataru (2017). Potential economic and environmental benefits from the interconnection of the Greek islands. International Journal of Global Warming. doi:10.1504/ijgw.2017.087211 · OpenAlex W4252968180, cited by 13.
  Interconnecting Greek islands saves EUR 9.7-17.8 billion over 2015-2040 in fuel and avoids EUR 1.8-2.6 billion of upgrades to island power stations: the closest cost-benefit analogue to linking oil-heavy Nusa Tenggara systems. See section.

1e. Benefits of a link beyond money saved

• new J. Kabouris et al. (2022). Power system static and dynamic security studies for the 1st phase of Crete Island Interconnection. Journal of Energy - Energija. doi:10.37798/201867278 · OpenAlex W4282832130, cited by 5.
  Crete (about 600 MW peak, oil-fired): the small system limits renewables and costs far more than the mainland; a 150 kV AC link is the first step. Mirrors Lombok and Sumbawa. See section.
• new John Vourdoubas (2023). The Interconnection of the Electric Grid in the Island of Crete, Greece, and its Contribution to the Clean Energy Transition. European Journal of Environment and Earth Sciences. doi:10.24018/ejgeo.2023.4.6.429 · OpenAlex W4390140228, cited by 4.
  After interconnection, Crete's cables can carry more than its peak demand, opening room for local solar and wind. See section.
• new Hannah Mareike Marczinkowski, Poul Alberg Østergaard and Søren Roth Djørup (2019). Transitioning Island Energy Systems—Local Conditions, Development Phases, and Renewable Energy Integration. Energies. doi:10.3390/en12183484 · OpenAlex W2972713547, cited by 49.
  Islands with different levels of interconnection reach high renewable shares in different ways; links and smart-grid measures are complements. See section.
• new Jon Sampedro et al. (2020). Quantifying the reductions in mortality from air-pollution by cancelling new coal power plants. Energy and Climate Change. doi:10.1016/j.egycc.2020.100023 · OpenAlex W3164747993, cited by 16.
  Cancelling all proposed coal plants avoids about 100,000-210,000 premature deaths in 2030 worldwide, mostly in developing Asia: the health side of not building new coal. See section.
• new Aman Malik et al. (2020). Reducing stranded assets through early action in the Indian power sector. Environmental Research Letters. doi:10.1088/1748-9326/ab8033 · OpenAlex W3012463309, cited by 50.
  India: continued coal expansion implies 133-237 GW of stranded capacity after 2030 under a well-below-2 C path; earlier action reduces it. See section.
• new Frederick van der Ploeg and Armon Rezai (2020). Stranded Assets in the Transition to a Carbon-Free Economy. Annual Review of Resource Economics. doi:10.1146/annurev-resource-110519-040938 · OpenAlex W2998548817, cited by 258.
  How stranded assets arise from policy and technology surprises, and who carries the risk. See section.
• new Andrea Tosatto et al. (2020). Sharing reserves through HVDC: Potential cost savings in the Nordic countries. IET Generation Transmission & Distribution. doi:10.1049/gtd2.12035 · OpenAlex W2996916932, cited by 25.
  Nordic case: using HVDC links for fast frequency reserves lowers the cost of keeping a low-inertia system secure. See section.
• new Pieter Tielens and Dirk Van Hertem (2015). The relevance of inertia in power systems. Renewable and Sustainable Energy Reviews. doi:10.1016/j.rser.2015.11.016 · OpenAlex W2183721398, cited by 797.
  Why inertia matters for frequency stability as systems take more inverter-based renewables; small island systems gain inertia from a link. See section.

Group 2. CGE and economy-wide modelling with the same aim

2a. Economy-wide models of interconnection, transmission and electricity trade

• in Appendix D Tianxiao Li, Pei Liu and Zheng Li (2019). Quantifying Cross-provincial Power Transmission Barriers in China, Based on A Computable General Equilibrium Approach. Energy Procedia. doi:10.1016/j.egypro.2019.01.136 · OpenAlex W2921452830, cited by 2.
  CGE of Sichuan hydropower sent to Jiangsu: the importing province's output and GHG both fall. The closest analogue to a Java import or export shock. See section.
• in Appendix D Jan Abrell and Sebastian Rausch (2016). Cross-Country Electricity Trade, Renewable Energy and European Transmission Infrastructure Policy. Journal of Environmental Economics and Management. doi:10.1016/j.jeem.2016.04.001 · OpenAlex W2300751791, cited by 142.
  General equilibrium with explicit transmission constraints for European electricity trade. See section.
• new Jian Wang, Shenghao Feng and Junyong Xiang (2021). Economic benefits of Northeast Asia energy interconnection: A quantitative analysis based on a computable general equilibrium model. Global Energy Interconnection. doi:10.1016/j.gloei.2021.07.002 · OpenAlex W3193555444, cited by 6.
  CGE of a Northeast Asia power interconnection: higher GDP with a trade-off between investment and consumption; estimates the CES between fossil and non-fossil generation bundles (cf. section 4.3). See section.
• new Alvaro Calzadilla et al. (2014). Desert Power 2050: Regional and sectoral impacts of renewable electricity production in Europe, the Middle East and North Africa. RePEc: Research Papers in Economics. · OpenAlex W145198141, cited by 8.
  Global CGE (DART) of exporting North African renewable power to Europe: gains depend on how plants are financed and on the cost of electricity. See section.
• new Budy Prasetyo Resosudarmo and Yuventus Effendi (2021). Implications of East Asia electricity market integration on Southeast Asian economies and CO2 emissions. Routledge eBooks. doi:10.4324/9781003190905-16 · OpenAlex W4200426998, cited by 1.
  Inter-regional SAM multipliers, microsimulation and CO2 for East Asian electricity market integration: some Southeast Asian economies gain, but regional CO2 rises. See section.
• new Peter G. WARR, Jayant Menon and Arief Anshory Yusuf (2009). Regional Economic Impacts of Cross-Border Infrastructure: A General Equilibrium Application to Thailand and Lao PDR. RePEc: Research Papers in Economics. · OpenAlex W2107677997, cited by 29.
  General equilibrium of a Thailand-Lao PDR cross-border link: short-run gains are small, long-run gains large once capital and labour move. Relevant to the closure choice in section 6.2. See section.

2b. Regional CGE and input-output models of electricity investment, and linking with power-system models

• new Mark Horridge (2012). The TERM Model and Its Database. Global issues in water policy. doi:10.1007/978-94-007-2876-9_2 · OpenAlex W2166456702, cited by 89.
  The TERM model and database, the parent of IndoTERM. See section.
• new Hongzhen Ni et al. (2022). Estimating the economic impact of large hydropower projects: a dynamic multi-regional computable general equilibrium analysis. Water Policy. doi:10.2166/wp.2022.119 · OpenAlex W4295970174, cited by 14.
  Dynamic multi-regional CGE (SinoTERM) with an electricity module substituting between generation types, used for large hydropower investment. See section.
• in Appendix D Arief Anshory Yusuf et al. (2023). Indonesian capital city relocation and regional economy's transition toward less carbon-intensive economy: An inter-regional CGE analysis. Japan and the World Economy. doi:10.1016/j.japwor.2023.101212 · OpenAlex W4386913471, cited by 22.
  Inter-regional CGE of Indonesia's capital relocation and the regional low-carbon transition. See section.
• new Albertus Girik Allo, Inayati Nuraini Dwiputri and Marcus R. Maspaitella (2022). The impact of electricity investment on inter-regional economic development in Indonesia: An Inter-Regional Input-Output (IRIO) approach. Journal of Socioeconomics and Development. doi:10.31328/jsed.v5i1.2775 · OpenAlex W4280529510, cited by 10.
  Inter-regional input-output model for Indonesia: electricity investment has a multiplier of 3.11 and spills over between regions; unelectrified villages are concentrated in eastern Indonesia. See section.
• in Appendix D Djoni Hartono et al. (2020). Phasing out Energy Subsidies to Improve Energy Mix: A Dead End. Energies. doi:10.3390/en13092281 · OpenAlex W3023074185, cited by 15.
  CGE of energy-subsidy phase-out and the energy mix in Indonesia. See section.
• in Appendix D Herbert Wibert Victor Hasudungan and Sulthon Sjahril Sabaruddin (2018). Financing Renewable Energy in Indonesia: A CGE Analysis of Feed-In Tariff Schemes. Bulletin of Indonesian Economic Studies. doi:10.1080/00074918.2018.1450961 · OpenAlex W2793657753, cited by 15.
  CGE of feed-in tariff financing for renewables in Indonesia. See section.
• new Anna Krook‐Riekkola et al. (2017). Challenges in top-down and bottom-up soft-linking: Lessons from linking a Swedish energy system model with a CGE model. Energy. doi:10.1016/j.energy.2017.09.107 · OpenAlex W2757043918, cited by 134.
  Lessons from soft-linking a CGE model with an energy-system model (TIMES-Sweden), with direction-specific connection points: relevant to linking IndoTERM-E with a dispatch model. See section.
• in Appendix D Jeffrey C Peters (2016). GTAP-E-Power: An Electricity-detailed Economy-wide Model. Journal of Global Economic Analysis. doi:10.21642/jgea.010204af · OpenAlex W2565090201, cited by 47.
  GTAP-E-Power: technology detail for electricity in an economy-wide model. See section.
• in Appendix D Yiyong Cai and Vipin Arora (2015). Disaggregating electricity generation technologies in CGE models: A revised technology bundle approach with an application to the U.S. Clean Power Plan. Applied Energy. doi:10.1016/j.apenergy.2015.05.041 · OpenAlex W1760483588, cited by 55.
  Technology bundle approach for generation technologies in CGE models. See section.
• in Appendix D Ian Sue Wing (2006). The synthesis of bottom-up and top-down approaches to climate policy modeling: Electric power technology detail in a social accounting framework. Energy Economics. doi:10.1016/j.eneco.2006.06.004 · OpenAlex W2026018101, cited by 152.
  Bottom-up electricity technology detail in a social accounting framework. See section.
• in Appendix D Qi Cui et al. (2020). Economic and climate impacts of reducing China's renewable electricity curtailment: A comparison between CGE models with alternative nesting structures of electricity. Energy Economics. doi:10.1016/j.eneco.2020.104892 · OpenAlex W3047062578, cited by 93.
  Compares CGE nesting structures for electricity (China, curtailment). See section.

Appendix E. Glossary

Appendix F. Command-file sketch for the first runs (not tested)

A starting point for the modelers, adapted from regdata/termener.cmf, regdata/LONGRUN.CMF and the dynamic files (COMMON.CLS, POL2.CLS, ALLINV*.PSH). Variable names are those of termener.tab; closures and set elements must be checked against the compiled model before use.

! ---- Static run, option A mechanics test (S1) ----
! atrad(COM,SRC,ORG,DST) is exogenous in the standard closure (termener.cmf, line 26)
Shock atrad("ElecGen","dom","JavaBali","Sumatera") = -10; ! delivered cost of Java power to Sumatera, percent
! ---- Option B: sourcing taste shift calibrated to a target TWh ----
Shock srctwist("ElecGen","dom","JavaBali","Sumatera") = <calibrated value>;
! ---- Idle Java capacity sensitivity ----
Shock acap("Elec_coal","JavaBali") = <surplus share, percent>; ! repeat for the other Java generators
! ---- Closure for a static run ----
! keep: swap phi = Natmacro("GDPPI"); swap xhouhtot = fhou; swap houslack = shrBoTnom; swap fgovtot = fgovtot3;
! remove: swap xcap = faccum; swap finv1 = finv4; Shock delunity = 1; Shock blabnat; Shock fxstocks; ashock delco2taxrate
! long run (see LONGRUN.CMF): fix rates of return (gret exogenous, xcap endogenous) and national employment;
! rename older variables to those of termener.tab
! ---- Dynamic run: policy file (.PSH), year-by-year ----
! Outer-island coal investment follows the plan instead of the rule (cf. POL2.CLS, which has this swap commented for renewables):
! swap xinvitot("Elec_coal",DST) = finv4("Elec_coal",DST);
tshock atrad("ElecGen","dom","JavaBali","Sumatera") = <yearly path>; ! ramp over the commissioning years
tshock xinvitot("Elec_coal","Sumatera") = <yearly percent change from the plan>;