How Indonesia can deliver rural toll roads

Earlier this week, BAPPENAS launched its priority list of PPP projects for 2015, including 5 prospective toll road projects and 1 potential toll road project. Following the tradition, they have included Manado – Bitung in their list of projects in the pipeline.

I wanted to write about this project because it illustrates a few problems with Indonesia’s approach to PPPs in the toll road sector, and one thing they’re (hopefully about to be) doing right.

History of the Manado – Bitung toll road project

The Manado – Bitung toll road is a 38km road planned to link the coastal cities of Manado and Bitung to support their integration into a unified metropolitan area across the province of North Sulawesi. In the 2013 PPP book, the investment cost was estimated at USD 353 million. I’ve never been to either of the cities, but I understand that this is a longer-term plan, and that the existing traffic flows, while heavy for the existing road, will not remotely cover the cost of construction and operation of the sort of toll road that has been proposed.

In the market soundings I have attended and briefings I have read, people spruiking the project have talked up the potential of the Bitung special economic zone (“SEZ”) and the upgrading of the port to drive traffic growth. There is some information on the SEZ here, and the upgrading of the port here.

The Manado – Bitung toll road has been included in every PPP book ever published by the Indonesian government. It was tendered in 2014, but the tender failed when it emerged that the winning bidder had no experience at all in the road sector and no ability to deliver the project (if I remember correctly, it was also tendered at some point in the late 2000s, but I can’t find a record of it).

Like all toll roads in Indonesia, the Manado – Bitung toll road was tendered with the private party taking full demand risk, getting all of their remuneration via user charges. The existing traffic flows were low, but government asked investors to take into account the potential traffic flows that might arise when the SEZ and upgraded port were finished. Given the uncertainty around the timelines of the development of these two facilities, and the impact they may have had on usage, all of the serious bidders took a conservative view of the traffic flows and realised that there was no way they would cover their costs through user charges alone. As there was no subsidy on offer, the serious bidders elected not to waste their time, leaving only poorly qualified bidders like the one that won.

So, all we need to do is offer a subsidy?

As I noted above, there is a lot of uncertainty surrounding the traffic flows. Private investors will necessarily take a conservative view on the traffic flows, but, with a sufficiently large subsidy, the project should be able to be made financially viable.

If the government wants the project, is that the best thing for them to do?

I would actually tend to think that, no, in fact, that would not be the best thing for government to do.

Excessive risk aversion transfers value to the private sector

The traffic flows for the toll road are highly uncertain. The government thinks that the traffic flows will be great, because they will develop the port and the SEZ, driving traffic, and they will create a conducive policy environment that will spur development along the route, further increasing traffic. Unfortunately for the government, the private sector bidders do not believe this, as the government has no track record of delivering projects like this on time.

The graph below shows, a simple depiction of what the two different scenarios could look like.

In a toll road, you get the vast majority of your revenues from user charges. Assuming you’ve got a single tariff for all cars regardless of distance, your revenues are the number of cars that drive on your road, multiplied by the tariff, as simple as that.

In an infrastructure deal, you’ve got two kinds of cash outflows: capital expenditure (“capex”), and operational expenditure (“opex”). Put very simply, capex is the investment you make upfront, which is fixed, and opex is the cost you bear over time, which changes depending on your traffic. If more people drive on your road, you have to spend more on maintenance. It’s a bit more complicated than that in the real world, but this will do for now.

The government looks at the projected costs and multiplies their traffic flows by the tariff; the difference between the two is the subsidy they think will cover the cost of service. The private sector party uses the same cost estimate, but with a more conservative estimate of traffic flows, calculate that they will need a relatively larger subsidy. In present value terms, the graph below shows how that arithmetic works with some simplified numbers.

In my simplified example, the government think they’ll need a USD 75 million subsidy, and the private party think they’ll need a USD 150 million subsidy.

Unfortunately for the government, what they think doesn’t matter. They run a tender with the minimum subsidy as the bid factor, so all the bidders get together and try and figure out ways to lower their costs, and raise their revenues, eventually arriving at the minimum subsidy they think they can get the job done for. Whoever bids the lowest subsidy wins, so the winning bidder’s expectation of the costs and traffic flows is what determines the subsidy, not the government’s. And, that’s the way it should be.

Through running a tender process, the government is trying to get to what economists call “efficiency.” The efficient subsidy will be an optimisation of the lowest costs capable of meeting the service standards and the best guess at the traffic flows. Each bidder will bid a subsidy that covers their costs, including a reasonable return on their investment, and no more. If they bid more, they would run the risk of being undercut by the other bidders. If the tender process is free and fair, you should come pretty close to the minimum subsidy possible to get the job done.

But, what happens if, in fact, the government exceeds the winning bidder’s expectation and delivers the port and the SEZ, and develops the area around the toll road in such a way that the government’s projection of the traffic flows were realised. The private party already got their subsidy, so the high traffic volumes just mean that the private sector makes a whole lot more money. They got a high subsidy, representing the conservative traffic forecast, then the high revenues from the actual high traffic volumes. The graph below shows what the cashflows could look like in present value terms.

In this simple example, the private party ends up making an economic profit of USD 45 million in present value terms. This means, they ended up covering their costs, including a return on their capital, plus a nice return over and above that.

How can a tender process result in economic profits?

But how can this happen? I said earlier that the subsidy was “efficient”, and efficiency means you just cover your costs, no economic profits. How can a free and fair tender process end up resulting in such a large economic profit?

For starters, in a project like this, someone will always make an economic profit or loss, because you will never predict exactly how many cars are going to drive on your road. If you get only one more car travelling on your road than you expected over the thirty year concession, the tariff paid by that additional car represents your economic profit, if you get one less, that’s your economic loss. In an uncertain world, efficiency means making the best possible guess with the best possible information available at any given time.

The winning bidder bid on the basis of the best information available at the time. They could not have known that the traffic flows would materialise in the way that they did. So, while, looking back, we can say they made a large economic profit, at the time of bid, the expected value of that profit was zero. Unfortunately for decision makers faced with uncertainty, as we live in the present, we’ll just have to live with the swings.

When something like this happens and a private party is making a whole lot of money, governments sometimes try to renegotiate the contract, to claw back some of the economic profit that the private sector is making. In the world of risk, that’s what we like to call a “heads I win, tails you lose” situation. Let’s say I offer to bet you USD 100 on a coin flip, if I win, I take your money, but if I lose, I try and renegotiate and say that I didn’t know it was going to turn out this way. This is analogous to what government is trying to do in some instances when they try and renegotiate contracts to claw back profits when things go well. If the private sector party over-estimated the traffic forecasts and was making an economic loss, the government would not (or, at least, should not) have stepped in to let them increase their tariff or to pay them extra subsidy, but when it is going well for the private party, they step in and try to renegotiate.*

What other options do we have?

At the other end of the demand risk allocation spectrum, you have what is called an availability payment, where the private party actually can have no exposure to traffic flows at all. As a simple example of an availability payment PPP, imagine a private party with a fixed monthly payment that they get as long as the road is in the minimum specified condition. This payment is set during the bid process to exactly cover their capex over a 30 year period. If they let the road condition slip either by building it shoddily in the first place, or by maintaining it poorly, they won’t get their initial investment back. On top of that, they have a separate payment that they get per car which is designed to cover their opex and no more. Revenue collection may be handled by the private party, but that goes straight into a government account.

The graph below shows a simple example of the cashflows in PV terms:

The biggest thing you may note is that the costs and revenues are different. This is because an availability PPP is seen as much lower risk to the private party than a full demand-risk PPP, so private parties can raise equity and borrow from banks at much lower cost, meaning they don’t need to ask for as much money to cover their debt service and repay their equity holders. But, on the other hand, as the government is bearing the risk, they should also discount the revenues they expect to receive to reflect the fact that they are bearing more risk.

I’m calling the subsidy an “effective subsidy” here because it’s not a payment in the same way that the subsidy for a demand risk subsidy is. The payments are the fixed and variable payments, the subsidy is just the residual between those payments and the revenues, whatever they may be.

As I mentioned earlier, this puts demand risk on the government. If the revenues are less than expected, the effective subsidy will be larger than expected, and the government will have to find more cash to pay the private party’s payments than it thought it would. If the revenues are higher than expected, the government takes the whole gain.

So, which is better?

In one case, the private party takes the risk, in the other, the government takes the risk. How do we decide which option is right for us?

True efficiency also means efficient risk allocation

The most basic principle of risk allocation is that you allocate the risk to the party best able to manage it**. If neither party has any control over the risk, then it doesn’t really matter which party you allocate the risk to. In a mature city, for example, there may not be much the government can do to influence traffic flows.

In this particular case, however, there are significant public actions that will need to be taken in the next few years, the government has much more control over the policy levers that will drive the traffic, while the private party will have almost no ability to influence demand. Further, the government has no track record of delivering the port, SEZ and surrounding infrastructure. Indonesia has over-promised and under-delivered in infrastructure for decades. No investor is going to bet hundreds of millions of dollars that the government will suddenly spring into action and deliver faster than it has ever delivered before.

Too many investors, and would-be investors have been burned trying to invest in Indonesian infrastructure, so private sector traffic projections for a toll road like this will always be conservative. If the government truly believes that it can deliver the work program, and it wants to get the value of the traffic flows associated with the delivery of that work program, then they must take on demand risk themselves. 

Government officials here are very risk averse (for good reason), so when you offer them a choice between bearing a risk themselves, or allocating it to the private sector, the initial reaction is almost always to dump it on the private sector party. But, if the risk is one that they are truly the best party to manage, they will end up paying more, and maybe even transferring value to the private party by forcing the issue. To me, that's a worse kerugian keuangan negara (state financial loss), than any loss government might make by taking on demand risk themselves.

Note: I am simplifying it a bit by just saying “government.” In fact, the contracting agency for toll roads is BPJT, the party responsible for building the port is Pelindo IV, the party responsible for building the SEZ is the provincial government of North Sulawesi, and the parties responsible for developing Manado and Bitung are the respective city governments. So, the Indonesian government as a whole has the control, but coordinating all of the parties will be a big job. BPJT alone won’t have the authority to knock heads together, but they’ll have a lot more access to the people that do than a private party. If you hire someone to build a road, get them to build a road, don't ask them to build a road, and coordinate 5 different government entities. That's government's job.

What is the Indonesian government doing to rectify this?

Earlier this year, the government issued Presidential Regulation 38/2015 concerning government cooperation with business entities in the provision of infrastructure. This replaces an earlier piece of legislation, and the new regulation explicitly incorporates language providing greater flexibility for the government to undertake availability-based PPPs.

So, will we see an availability payment being applied for Manado – Bitung?

It’s too early to say. I hope so. I suspect it’s probably appropriate for a toll road like Balikpapan – Samarinda as well.

There has been talk about the use of performance-based annuity schemes (“PBAS”, which is just a fancy name for an availability payment) for the construction of non-tolled roads, but I haven’t yet seen any statement from BPJT on whether they intend to use this model for tolled roads as well. I guess we’ll see what they think when BAPPENAS finally publishes the PPP book with more information.

*Note that the proscription on renegotiating does not apply where projects were not competitively awarded. This happens a lot in resource projects in developing countries either due to information asymmetries, or straight out corruption (the New Yorker has an example of one here). Renegotiation or even termination could be the efficient outcome there.

**If you want to know more, a good treatment of this is in Tim Irwin’s book Government Guarantees, available in PDF here. Aside from the discussion of risk, it’s a fantastic resource for people in the infrastructure business.

Jakarta taxis need peak pricing

I took a Kopaja bus from Polda Metro to Blok M during rush hour the other day because it was starting to rain and I couldn't find a cab. I got to Blok M and wandered over to SMAN 6, where taxi drivers often hang out. Like usual, there were 5-6 taxis parked there, all from good companies, but all seemed like they were eating or sleeping, so I wandered a bit further. 

As I walked, two available Blue Bird taxis drove past me before one stopped. He said he was actually on his way to take a rest, but he'd take me because I was close. 

I was confused: why would taxis be choosing to rest during peak times? Surely that's when it's easiest to get a passenger. 

The taxi driver said, yes, it is easiest to get a passenger during peak times, but you have to sit in traffic longer to get to them, the meter runs slow because of the heavy traffic, and, if you get someone that takes you to a really out of the way place, you can get stuck in traffic trying to get back to a place where you can get your next passenger.  In this sort of situation, as a taxi driver, you're better off waiting out the peak, and jumping back in your cab when the traffic dies down.  

If you work in Jakarta, you're familiar with the queues for taxis at peak times. At certain buildings like the Stock Exchange Building or Sampoerna Strategic Square, an hour's queue at 6pm is pretty standard, two hours is not unheard of. 

The taxi queue times have negative impacts on society. Massive amounts of time is wasted that citizens could put to better use, and some people will choose to buy private cars further adding to congestion. On the plus side, some people will choose to take buses, but the length of the taxi queues suggests that there's a significant population that won't*. 

Peak pricing, or allowing taxis to charge more during peak times, would have two major positive impacts. First, it would encourage drivers that currently sit out the peak to come back on to the road. Secondly, as drivers and taxi companies would be making more money, they would be willing to put even more taxis on the road. Both of these things would decrease queuing times, making riding a taxi relatively more attractive compared to private cars.

On the negative side, there would be people that wouldn't be able to afford the higher tariff, but those people could still ride buses, or choose to travel before or after the peak times.  

So, if all these good things follow from peak pricing, why haven't we got it already? My guess would be that it's because companies are scared to be the first one to do it. It might be more economically efficient, but it's usually not popular among your customers. Being the first company to do it would most likely mean you would lose customers to your competition.

It's also unclear what the provincial government of DKI Jakarta would have to say about this. My understanding of taxi fare regulation is that it is governed by Perda 12/2003, pasal 77 of which says that taxi companies are allowed to set their own tariffs, but they must be approved by the Governor**. It's not impossible that if a taxi company did try to implement peak pricing, someone within the Jakarta government might think they could score political points by opposing it. 

As I said earlier, I think there are benefits to broader society from peak pricing that would accrue in the form of less congestion. Unlike shorter waiting times for customers, and higher profits to the taxi companies, the lower congestion is a positive externality that accrues to society from these two parties undertaking the transaction. 

On the basis of this externality, I think there's a role for DKI to not just not oppose peak pricing, but to encourage taxi companies to take it up. If DKI Jakarta were to publicly encourage taxi companies to adopt peak pricing and to talk about the broader benefits to society, citizens would be less likely to react poorly. 

I, for one, hope they do soon!

*I've actually been in both the populations that buy cars in response to taxi queues and those that ride buses. I would be using my private car now, except for my own peak demand problem, where my wife and daughter want to use the car in the morning and evening. Lowest value users get sidelined... Buses and taxis for me. 

**Note that this is the same regulation under which various people are complaining that Uber is illegal, because it is charging "illegal" tariffs that haven't been approved by the governor.