By Angus McCrone
Bloomberg New Energy Finance
We do not hear very much about “green jobs” these days. Around the turn of the decade, politicians were fond of putting those two words together and making grand claims about the number they would create – five million in the case of Barack Obama during the 2008 presidential campaign.
Now, the talk is more about brown jobs, even if no one actually uses that term. The likes of Donald Trump in the U.S. and Beata Szydlo, prime minister of Poland, talk about protecting coal miners, and say little or nothing about the employment opportunities in clean energy.
There is logic to the change in rhetoric. When a sector is under pressure – as renewables were after the financial crisis, or coal and oil are now – it is not surprising that its supporters resort to claims about jobs. When a sector is winning, they do not need to.
This shift in the political weather from trumpeting green jobs to chest-beating about brown jobs leads us onto something that deserves more attention than it has got so far. That is the wider effect on economies around the world of the transition to clean energy and transport.
In power generation, we are moving from centralized power stations burning raw materials extracted from below ground, to distributed units harnessing natural resources, and in transport, we are starting to move from the internal combustion engine based on liquid fuels, to electrified vehicles.
This article will look at the economic implications, using BNEF’s own long-term forecasts – the New Energy Outlook 2017, known as NEO, published in June this year; and the Long-Term Electric Vehicle Outlook 2017, published in July. In energy, NEO 2017 predicts that wind and solar will account for 34% of world electricity generation by 2040, compared to 5% now. In transport, BNEF’s forecast for the light-duty vehicle market worldwide projects that 54% of new car sales will be electric by 2040, the result of remorseless reductions in battery costs, increasingly stringent regulations, and growing commitments from automakers.
Two things before we start: many of the economic effects postulated below would still apply, albeit not as dramatically or as quickly, if the future turned out to be closer to that portrayed by the forecasting models of the International Energy Agency or one of the big oil companies. And what follows is a first stab at a topic that is highly complex, with innumerable moving parts. I would welcome feedback – and contrary views!
Let’s start with economic activity, or GDP, and move on to look in turn at inflation, international trade, jobs and government finances.
In much of the clean energy transition, the world is moving from technologies in which lifetime costs are split between upfront capital expenditure and ongoing fuel purchasing, to technologies in which upfront capex is dominant.
I am going to concentrate on wind, solar and electric vehicles, although the same point would also apply with expensive upfront, but cheap-to-run, consumer energy appliances such as LEDs.
For wind and solar, I see two phases of the impact on GDP. The first is when the new technologies are been supported by government subsidies; the second is when those technologies are less expensive than any available fossil fuel alternative on a lifetime cost basis but when upfront costs are still higher.
Let’s take the first phase – roughly where we have been for most of the last decade with wind and solar. In that era, there is a boost to GDP as capital expenditure to produce a given amount of electricity (let’s say MWh per year) runs at a significantly higher level than if generators had chosen to produce those megawatt-hours instead from new fossil fuel plants.
The capex differential has shrunk over time, but still exists. In Germany, for instance, our estimates are that a utility-scale PV plant without tracking could be built in the first half of 2017 for about $950,000 per MW, an onshore wind farm for $1.7 million per MW, an offshore wind array for $4.1 million, a coal-fired plant for $1.6 million and a CCGT plant for $930,000. Then you have to adjust for the relatively low capacity factors for wind and solar (say 25-50% for wind, 10-25% for PV), and that makes the capex for those technologies far higher per MWh per year.
Phase one, the era of subsidized wind and solar, has a lingering negative effect on GDP, as well as the positive capex one discussed above. That is because, during the period of the subsidy, the resulting electricity will be more expensive for consumers and businesses, than if generators had gone for new coal or gas plants. There may be a benefit to the planet from lower carbon emissions, but that is not included in GDP.
Many countries will soon embark on phase two, and a few have already. Phase two is the period when wind and solar are still more expensive upfront but, on a lifetime (or levelized cost of electricity, LCOE) basis, they are cheaper than coal or gas, let alone nuclear. In the first half of 2017, according to BNEF analysis, either wind or solar was the cheaper as a new generation source on an LCOE basis than any fossil fuel option in Brazil, Chile, Australia, the U.K. and Germany.
Phase two sees a double boost to GDP because generators are choosing to deploy more capex upfront for a given number of MWh per year, and because the resulting electricity will be cheaper for consumers and businesses than if they had instead built conventional power stations. Even countries that locked in extra electricity cost as a result of phase-one subsidies will eventually (sometime in the 2025-2035 period) see that impact on power prices fade out.
There are two other complicating factors on energy and GDP that I will leave to the reader to weigh up. The first is that, theoretically, extra capex on new power generation equipment could “crowd out” other investment in the economy by pushing up borrowing costs, and that could offset its positive effect on GDP. But, given that interest rates are at previously undreamt-of lows, that seems unlikely for now.
The second is the fact that increasing use of variable sources such as wind and solar will require investment in balancing capacity in order to ensure that electricity supply always meets demand. The amount of balancing needed will vary by country, and there will be a range of different options available. These will include interconnectors, new quick ramp-up gas-fired plants, batteries and demand response. Some of these will boost GDP as a result of upfront capex but may reduce it over time if repaying their capital cost involves adding to the overall cost of electricity.
In transport, different phases will also be important. Between now and the 2030s, according to BNEF forecasts, EVs will ‘cross over’ internal combustion engine, or ICE, cars first in terms of lifetime costs and later in terms of upfront purchase prices too. This will happen mainly because of further deflation in the cost of batteries.
At present, the upfront cost of a medium-sized EV in the U.S. is about $44,000, against around $28,000 for an equivalent average gasoline or diesel vehicle. So decisions to buy EVs rather than ICE cars will involve higher expenditure upfront by the consumer. That should raise the consumers’ expenditure component of GDP, provided that the extra spending comes out of money that would otherwise be saved and therefore is not offset by lower spending on other items.
Phase two in transport (roughly from the late 2020s onwards) will see EVs starting to undercut ICE cars not just in terms of lifetime costs but also in terms of their upfront purchase prices. On the face of it, that should result in lower consumer spending, and therefore lower GDP growth, than if the same motorists were buying conventional cars.
In practice, that negative effect might be delayed, and it might be largely offset. Delayed, because the electric engine technology developed in passenger cars would spread to other areas of transport such as trucks, sparking a rush of spending there, sometime after the take-off for EVs in the light-duty car and bus segments. There would also be a build-out of EV charging infrastructure, boosting fixed investment. And largely offset, because much of the gain for consumers from being able to buy cheaper vehicles could be recycled into spending on other things.
Inflation reduced long-term
One link between clean energy and consumer prices has taken almost all the attention of media and public in recent years. That is the impact “green charges” resulting from feed-in tariffs and green certificates have on electricity prices, and therefore inflation. This impact has been controversial. Subsidies amount to an estimated 13% of current U.K. electricity bills, for instance, and 24% of those in Germany.
The overall inflationary impulse resulting from subsidies may be small – electricity has a weight of just 17 out of 1,000 in the U.K.’s Consumer Price Index. But it will continue to be felt in the countries concerned for a while, particularly if it feeds through to the prices that electricity-consuming industries charge for their products. It will go into reverse when the amount of subsidy attached to existing projects reaching the end of their feed-in tariff or green certificate period outweighs the amount of subsidy awarded to new projects.
In other countries that have not subsidized green power via electricity bills, renewables will start to exert a downward force on inflation once power purchase agreements for wind and solar are clearly below the prices offered by developers of other technologies. This point has already been reached in some locations, and it will do so in many more, assuming that the LCOE of wind and solar continue to fall as we expect. For instance, BNEF’s New Energy Outlook forecasts that the average LCOE of utility-scale photovoltaics in the U.S. will fall from $64 per megawatt-hour in 2017, to $19 in 2040. In some areas, clean energy has already had a soothing influence on bills, by providing a glut of power at times of day that used to see peak prices.
EVs should also exert downward pressure on inflation, as they become a significant part of the car market and their prices fall in the late 2020s and 2030s. And cheaper electricity and cheaper cars in the medium term would be a restraining influence on global interest rates.
Seismic shifts in trade
International trade stands to be affected radically by the clean energy transition, via the supply and demand for commodities and equipment.
An obvious pinch will be felt by exporters of coal. BNEF’s NEO forecast projects only a 5% fall in coal-fired generation globally out to 2040, but the nature of that will change – with a big rise in the use of domestically mined coal in India and South East Asia, and less use of seaborne coal by China and, in particular, the developed countries of the OECD. Rather than being a growth market, as it has been since 1990, internationally traded coal is going to be in decline from a peak in the 2020s – with knock-on negative effects for exporters such as Australia, Indonesia and Russia. In practice, weaker demand would put pressure on prices, hammering the more expensive producers the hardest.
Unlike coal, gas is forecast to grow as a generation fuel. Overall, BNEF predicts that gas-fired generation globally will rise 10% between 2016 and 2040, with the biggest increases in the Middle East and Africa and the U.S. The U.S. trade balance has started to benefit from LNG exports from giant terminals such as Sabine Pass in Texas, and our analysts expect global LNG trade to expand at least for the next 10 years.
As for oil, BNEF’s EV forecast envisages the rise of electric cars taking out demand for 8 million barrels per day of transport fuel equivalent by 2040, compared to what it would have been if all cars on the road at that date were efficient ICE models. Any erosion of demand for oil between now and 2040 is negative for crude exporters such as Saudi Arabia, Venezuela and Russia, already under pressure from North American shale oil.
What about the new technologies? On the commodity side, the energy transition should be a boon for suppliers of the key raw materials for solar and for electric vehicles. This should benefit the export earnings of China (for rare earth metals), Australia, Chile and Argentina (for lithium) and Democratic Republic of Congo (for cobalt, also a vital ingredient for batteries). Of course, “benefit” is only the right word here if the country concerned has the governance to cope with it, and that is a particular issue for the DRC.
Exports and imports of batteries will also affect countries’ trade balances. The aggregate volume of lithium-ion batteries sold worldwide (for EVs) is expected to increase 40-fold between 2016 and 2030 on our forecasts.
The supply of these is likely to be concentrated in a few countries. A Research Note earlier this year by my colleague I-Chun Hsiao reported that Chinese-headquartered companies already have, or are planning, 159GWh-a-year of lithium-ion battery capacity, compared to 48GWh-a-year for Korean firms, 39GWh-a-year for U.S. corporations and 27GWh-a-year for Japanese players. The rest of the world is far, far behind. Clients can read that note here or on the Bloomberg Terminal here.
Even if the likes of Germany, the U.S. and France export the same number of autos in 2040 as they do today, their net export earnings will be significantly less if the batteries in those vehicles have to be imported from the Far East.
On wind power, NEO forecasts that annual capacity additions will increase from the 60-gigawatt-a-year run rate seen recently to about 140 gigawatts per year by 2040, including the net contribution from repowering old plants. This provides an export opportunity for turbine-making centers such as China, Germany and Denmark.
Countries relying on imported wind equipment will see trade balances affected during the ramp-up period. For instance, the U.K. is adding about 1.2GW a year of offshore wind at the moment, costing in capex some $4.5 million per MW, or $5.4 billion in total per year. If half of that expense is made up of imported turbines, then the U.K. trade balance is currently laboring under a $2.7 billion-a-year burden.
In solar, BNEF predicts an increase in the annual rate of global PV additions from 76 gigawatts in 2016 to 300 gigawatts per year in the late 2030s. Even though prices for solar modules will continue to fall at an estimated learning rate of 28%, according to our estimates, the real-terms dollar value of the PV market in 2040 is likely to be 1.75 times that of last year – providing an export boost for China and any other country that becomes a hub for polysilicon making and module manufacture. However, countries forecast to add a lot of PV capacity but unlikely to be manufacturing centers, such as in Southern Asia, the Middle East, Europe and Africa, can expect to see higher imports during the ramp-up period.
Once the ramp-up phase peaks, countries with new wind and solar plants should enjoy a boost to trade balances, if that clean electricity bites into imports of fossil fuel commodities such as coal and gas. Countries with expected high renewables penetration, including Germany, the U.K., Italy, Japan, and Chile, and a number of smaller developing economies with no domestic fossil fuel extraction, could enjoy that benefit.
The outlook on employment is closely related to that for GDP and trade. In power generation, coal-firing worldwide ceases to grow according to our forecast, with increases in South Asia more than offset by declines in developed economies. This will mean a shrinkage in mining jobs in Europe and the U.S., and in countries that produce a lot of seaborne coal such as Australia.
In renewables, our projections for year-on-year additions suggest that employment in both wind turbine and solar panel manufacturing globally will continue to grow at least until the late 2030s, as capacity additions jump from a combined 636GW in the 2021-2015 period, to 2.2TW in 2036-2040. However, the rate of job creation growth will be tempered by rapid improvements in productivity – the flip side of the sharp cost reductions NEO predicts.
There will also be increases in the number of people involved in constructing wind farms and utility-scale solar projects, again mainly in developing countries. As an indicator, Adani Power’s 648MW Kamuthi PV park in Tamil Nadu employed 8,500 workers for six months in the building phase in 2015-2016 – equivalent to 6.6 man-years per megawatt. The latter figure is roughly double what might be expected in developed economies. Even if productivity in construction improves, the solar (and wind) project build-out globally should keep many hundreds of thousands of hard hats busy through the 2030s.
And then there is small-scale solar system installation. This is a much more labor-intensive business than polysilicon, wafer, cell, module or inverter manufacturing, or indeed the building of big PV parks. Estimates are that 11-13 man-years are required to install 1MW of small-scale solar, although it does depend greatly on whether the systems are household rooftops of 3kW or so, or commercial-scale arrays of many tens of kilowatts. The majority of the new jobs will be in developing countries (since these are forecast to add more than 700GW of the 1.3TW of new small-scale PV installed worldwide between 2017 and 2040).
In the operating phase, there are many fewer jobs per MWh in wind and solar plants than there are in fossil-fuel generation. For instance, the giant 2GW, $4.5 billion onshore wind project in the Oklahoma Panhandle announced this summer by American Electric Power, is slated to create 4,000 jobs during the construction phase but only 80 during operation (a ratio of just 0.04 per MW). By comparison, AEP’s 600MW John W Turk Jr coal-fired power station in Arkansas employs 109 people directly, plus those in the coal mining and transport chain, to produce about half, at most, of the electricity that is due to come from the Oklahoma wind complex.
Transport may see the most striking changes in employment. A VIP Comment article in August 2016 by BNEF founder Michael Liebreich and myself, entitled Electric Vehicles – It’s Not Just About the Car, argued that because electric engines are much simpler than ICEs, the amount spent on servicing will be much lower. Visits to the dealer to replace brake pads and tires will still be required, but oil changes will not, and neither will new spark plugs. For as long as there remain large numbers of ICE cars on the road, the blow to jobs in servicing – and in gasoline and diesel retailing – will be softened. But the advent of EVs will mean many fewer employed in these areas than if conventional cars continued to be nearly 100% of the market.
Manufacturing will also change. The number of moving parts in a car engine is much lower for an electric model than for an ICE vehicle, with a recent report by UBS putting the powertrain of the Chevrolet Bolt EV at 24 moving parts compared to 149 for that of a conventional car. This is likely to mean pressure on car manufacturing workforces in countries like Germany, Japan, South Korea, the U.S. and China.
ICE cars will take a long time to disappear: BNEF’s forecast has 55.6 million ICE vehicles still being sold globally in 2040, more than two thirds of last year’s total. However, the advance in robotics will mean that car manufacturing will continue to become less labor-intensive between now and 2040.
Then there is the whole dimension of autonomous driving. The job losses from this technology could, in theory, be catastrophic – Bureau of Labor Statistics data show that there were 665,000 bus drivers in the U.S. in 2014, 2.7 million truck and delivery drivers, and 234,000 taxi drivers and chauffeurs. The likely speed of autonomous driving take-up worldwide, and the spread of car sharing and ride-hailing, is the subject of much debate at present, with the impact potentially being very significant on consumer spending and GDP and well as jobs. BNEF will be publishing more research on this topic, from a new team of analysts covering “intelligent mobility”.
A hole in government finances
Many developed economies tax road fuel heavily. For instance, the U.K. government pocketed 28 billion pounds from road fuel duties in 2016-2017, plus some extra billions from value added tax on the same fuel. The combined total was enough to pay for 4% of total public expenditure in that year. Germany took 36.7 billion euros in tax on gasoline and diesel in 2016, equivalent (even without the VAT contribution) to 3% of its public spending. Even those countries that tax it more lightly, including the U.S., nevertheless collect significant amounts of sales tax on gasoline dispensed at the pump. Not all states levy sales tax but if we take 6% as a rough average, and gasoline sales at $983 billion in 2016, then the U.S. tax take may have been around $60 billion.
As road transport switches towards electric in the 2020s and 2030s, developed economies are going to face the gradual loss of that road fuel tax. Our advanced transport team predicts that by 2040, the number of light-duty cars on the road in North America and Europe that are conventionally powered will be 181 million, down from 249 million now. As if that is not difficult enough for treasuries to deal with, fuel efficiency in ICE vehicles is likely to advance steadily between now and 2040, as mounting competition from EVs spurs manufacturers to reduce further the running costs of conventional cars, buses and trucks.
The lost tax revenue from road fuel will have to be made up via other imposts, affecting EV drivers as well as ICE owners. The obvious candidate is some sort of charge related to the number of miles driven on the road, perhaps skewed to a higher fee for congested routes or higher polluting vehicles. In July, London-based forecasting house CEBR proposed a system in which fossil-fuel-free and conventional motorists would pay road usage charges varying with congestion, effectively a rent for the land use involved, amounting to more than 30 billion pounds a year by 2037. This would be on top of the residual take from current taxes, and the result would be more money going into road improvements.
As I warned upfront, this has been a first, imperfect look into a massive topic involving trillions of dollars, and multi-decadal shifts.
What is clear is that the clean energy transition will pose tough questions for those in charge of the world’s economies. Two of the most difficult will be how oil-exporting states can survive their loss of export earnings, and how some of the world’s most successful industrial economies can cope with a loss of car manufacturing jobs. Generally, insofar as capital-intensive technologies substitute for more labor-intensive ones, there could be uncomfortable implications for income and wealth distribution.
There will be big business opportunities too, not just in the activities mentioned above but in other corners of the clean energy transition, such as stationary storage, grid and balancing services, and the installation and operation of mini-grids. Software development and the “internet of things” will make possible the collection and exchange of data for the coordination of devices in the grid, in the home and in the workplace. Many of the new activities will be knowledge-based and much less dangerous and dirty than coal-mining or oil-drilling.
(Corrections were made on September 5 to the paragraph about trade in gas, and to the paragraph about road usage charging.)