I know we're, a little bit late so if, if you want to stay through the whole lecture that's great otherwise I can understand if you leave a little bit early. In fact I was encouraged to do this lecture not simply by Dr Chang but also my sons. I have five sons who graduated from Princeton, so, so this is like old home week. Anyway, I'm sure, they encouraged me and I suspect after they see this lecture they'll probably give me some constructive criticism. Anyway, because of the time, I have a lot of slides, it's going to take about 30 minutes, Even with the 30 minutes, it will be a little bit rushed. So there are a lot of notes at the end of the slides. If you want to so if you want more details, you can access them there. So, as Professor Chang said, the National Exchange Carrier Association. Sets access rates, I'll define those in a few minutes. For more then 1100, all, all most 1100 rural telephone companies. And I suspect most people in this room don't realize that there are over 1100 telephone companies out there, but there are. Our companies serve. Over 40% of the geographic area of the United States, but only have five million subscribers. So, very, very sparse territories. And these are the areas that our members, serve. So you can see it's midwest, far west, and Alaska. And if you superimposed Alaska over the 48 states, you'd see it's about a third of that territory. So it's really huge areas. Very isolated communities. So what do I mean by an access rate? It's just a wholesale rate. Charge to retail providers of communication services. So for example traditional rates were switched to access rates. And those were charged. To long distance carriers like AT&T for completing long distance calls over these local networks. They don't, AT&T doesn't have a network that goes to each individual customer. So it, our members would charge them for originating or terminating a long distance call. On the local network. There are other types of rates called special access rates, they're mainly data associated rates or dedicated networks. So for example, internet service providers. Buy from the local telephone company Digital Subscriber Loop (DSL) service. So that's the wholesale service for internet service providers. Wireless carriers, you may think that almost all of that traffic is going over the air. Well it does for a small segment but actually to back all that traffic to a mobile switching office and then farther into that network, actually wireless carriers buy a lot of land line facilities to connect their towers to mobile switching offices, for example. And finally, there are private. Fit net, private line network providers. Some people don't want to use public network. They want their own private network. Now, what are the demand trends for the traditional voice service, traditional long distance service that you're familiar with. Well, that's a dying service. The connections to the Internet are flattening out in rural areas. But broadband services, especially wireless companies are growing tremendously. And, although digital subscriber line services, demand for it is flattening, it doesn't mean that the amount of bandwidth used for, the Internet is flattening also. These, the customers are using, are accessing video services now. So, here's some charts to at least give you a sense of what's, what's happening in the marketplace. And rural companies really are indicative of the bigger companies also. Now, what I have here is annualized growth rate. So if you take let's say, June of 2011. And divide by June of 2010, you can get an annualized growth rate. If you look at traditional long distance service, you'll see that it's, it's declining by about ten percent per year. So it's, it's pretty substantial and it's been going on for quite some time. It's a, a legacy service. Data digital subscriber line growth is actually been pretty strong, but it's reached a mature stage and it's only growing at about four percent per year. So, flattening out. The big growth area, now this is all special access, that dedicated data type of traffic. That includes digital subscriber lines. Includes, wireless backhaul and so on. This is growing at, a relatively fast pace still. At about eight percent to ten percent per year. And it tells you that wireless carriers need to buy special facilities to link their, their network, their towers to their mobile switching offices have, because of 3G, 4G and all these other applications, they're buying facilities at a rapid pace. So this isn't exactly a strictly correct definition, software defined networks. I just want to get it, get across the idea that there are networks that are basically defined virtually. And some of them have special transportation routes. Sometimes there's a tunneling of, of, of band width for particular customers. And you can think of them as, bright, broke breaking up into public broadband networks like the Internet. There's a priv- there are private bo- broadband networks. And that's typically not a cross-, Right now it's, it's in its infancy, infancy. Most large carriers will provide a virtual private network to their customers, But it's not that easy to go from one network to the next because the standards for transferring data from transferring packets. From one network to another is, is, still being worked out. And then there are hybrid networks. These, private clouds, public clouds. And then, a private network often will, have, access to the Internet. So that's a, a hybrid network. And there are still physically separate networks. Some, some types of, customers, do not want to use, facilities that anybody else can use. Expensive, but it's there. Now, when we get to network pricing. I want to. I, you've seen some mathematical models this morning. I want you to understand that pricing is really an art and a science. You're really, when you're trying to price, you're in this kinda zone of indeterminacy. You really don't know what's happened in the market yesterday. You probably don't know exactly what's going on in the market today and you certainly don't know very well what's going to happen in next months or years. So you, you're, in that zone of indeterminacy. There is a lot of reasonable solution to pricing. And one thing that we've decided is, it's, and you heard it earlier is that you want to keep your pricing strategy simple, unless there's a big payoff for not being simple. So, for example, price could be set equal to unit cost for a, a particular service. Now, you don't want to set it you may want to adjust that, be a little bit more flexible if some people willing to pay a little bit extra you know, luxury customers. So, keep it simple and the pricing itself. If it's simple, people understand. Second is you want to tap market information. If you're setting a price. Okay. If you're setting a price well, you're not setting it an isolation. There are many other, there many other. Entities out there that are setting prices also, you might as well piggyback on it, however they may face different market conditions than you do, and you may not agree with their pricing strategy. For example the economic models treat customers as extremely rational. They and they actually decide with a lot of information over a long time horizon. It could be an infinite time horizon. That's not, that's not the case in the actual market. You may, you may, for example, see a lot of empirical work on what the relation, people, people may demand based on the price in the market and some other demographic factors. You have to keep in mind also that most of these models. I'm not terribly stable. You may get a relationship and it looks good. And then you look at it three years from now and it, and it, and it no longer works. And you'll find that typically that these empirical models assume a, a, a stability in the market that often isn't there. And rely and they often rely on the to test the significance of relationships on what's know as asymptotic theory. Where the distribution becomes stable. Usually a chi square as time goes to infinity. And chances are you've seen that if you've, if you've watched the options market or the financial meltdown, that a lot of these models are not. They assume there's a stability that isn't there. So prices should also, so. Science is good because it does help you with all those caveats, it does help you structure your inquiry because these models do highlight important factors that you ought to consider in pricing. And the empirical work can identify some basic challenges. So I'm not saying that's it, science is quite important in trying to set a pricing strategy, but don't treat it as if it's mechanical, okay? Mechanical engineering, or something along those lines. Now, you should also think of price in context. It's a mark-, it generates revenue, but it's also a marketing tool. How do I get it back? [inaudible] Huh? It's back. Okay. So. So let me go on and, and here are the three areas that I'll talk about in depth is our basic pricing model. Implementation issues associated with the pricing model. And I'll give you base a simple example from DSL, digital subscriber line, how we price. Okay? So one thing to keep in mind is the basing pricing model that we use. Now you've focused on demand today. Well, there's the supply side also. And you'll find in economics textbooks that this kind of model isn't at odds with terribly at odds with what you'll find in an economics textbook for long run pricing. The top, the top model, the top equation just as the price of this surface must equal its operating cost. Plus, now that's direct investment of, in the service the particular equipment associated. You multiply it by r which is a targeted rate of return, which turns it into a rental price, and you divide it by the amount of units that you, you're selling. So basically you have operating cost plus rental cost associated with the service divided by the units and that gives you a. Unit. Direct cause for this service and then you raise it by a markup, and the markup could be depend on market conditions. Okay, And that could take into the demand side. It could take into account other types of factors. Because this is a mature industry you use this for a particular service but for all services you want revenue and total from all of your services to equal the total operating costs that you plus, Plus the rental cost of your equipment in total. Plus the rental price of your overhead and what do I mean by overhead. So direct investment is in, let's say, the multiplexers associated with a particular service any type of server equipment. Overhead is really buildings, Cars and so on. You have to, you have to recover those costs. Also, you have to pay back your investors and you have to pay off loans. So your revenue, if you're in a mature market, ought to ought to generate enough to cover your costs, including a return on investment. So let, let's now, even that simple model. Is too difficult in a lot of cases because, a lot of times you can't associate. You have a couple of workers out in the field. And the operating costs associated with a particular service is not easy to pin down. So what we do is, we look at the whole income statement of a company. And we look at the total direct investment, the total operating costs of a company. And we get a factor called B. So once we do that, we link operating costs for a particular service, to a direct investment in a particular service. So if you go back, you'll see that we had operating costs plus rental price. Now we're going to link price completely to equipment by making this connection with operating costs. And the, we call this a direct cost factor. And so r is the required rate of return for investors and then you add a little bit to cover operating costs. Okay, that's the way to understand it. So let me talk about some of the implementation issues. You know, it seems like a simple model. Remember, we're talking about wholesale services that are sold to companies' that are trying to produce retail communication services. First of all, we have to define the service. Second, we have to figure out what we mean by that R, that targeted rate of return. Third, we have to talk about what the unit cost is, and finally the markup. And none of them are particularly. I'm going to go through all the complications, and then make it simple at the end, to give you a sense of what's going on. So first of all, what is the service that you're providing on a wholesale basis. Well, you I think these should be familiar. Upload and download speeds. Reliability of connection and throughput. You can also have some, I should have put in special switching and routing to improve class of service. Class of service is just prioritizing packets. Quality of service is to improve these, the smooth streaming of it so you don't have jitter when you look at television sets. Security and privacy. So, those are some of the features, some of these are, are, are hard to measure their value. And the other, the other kind of emerging way of looking at, at our wholesale services is, is to look at our network as a platform. And there are two sides. To a platform, it's the end user side and the other side where they're getting access to let's say websites. And it, this platform approach, for example, I guess most concrete example is, if you own a singles bar, that's the platform. Men and women come in. There are the two sides of the market, And then the question is, who should you charge? The men, or the women? And typically you'd charge the men an entrance fee into the into the platform, the bar, And women go free. So who, who, who pays for some of these services? You could look at it from a platform approach. Alright. So, I've defined the service. It's bandwidth, quality of service, class of service, and so on. It could be other types of features of a smart network. If you remember, the basic pricing model had direct investment times r to make it a rental price. Well, what is that targeted rate of return? And I won't go into it. I put some notes onto these slides. But one thing that you should notice is that it's. There are a lot of financial theories out there. You can open up textbooks, and they look really fancy. There are all sorts of models out there. In general, they don't work too well. Again, its an art, more than a science. And that's why we've had financial meltdowns. Let me give you a concrete example. Which is the right, eh, interest rate to ass, to apply to a rural telephone net, company's network if they're borrowing. Is it a mortgage rate at four percent or a credit card debt rate at fifteen%. A mortgage rate of four percent is usually low, because you make a big down payment. You can sell off the house. There's a, a lot of screening of credit, the customer. With credit card debt, there isn't a, there, there's isn't a, a down payment. There isn't a, a house there. Think about a rural telephone company. Suppose it goes bankrupt. What can you sell? The copper in the ground or the fiber in the ground, if it's not doing well. So the question is, which is the right rate cost of debt. And I could go through the cost of equity and so on. The answer is, it's by no means clear. And you can, none of these theories are, if you, if you ever take financial investment theory or option theory, the models have their challenges. So next is, well okay, let's look at unit cost. Now we've linked everything to unit cost to direct investment. So what is the direct investment? There's loop and it could be fiber, copper, fixed wireless. Each one has a different cost structure. You buy it for a different price. There are ports that have speeds and features equipment. There's a transport backbone. For, for rooting in distance and the, the same transport backbone. It's used by many services so how do you split that out? And then there're special features. They're they'd be usually built-in to the port. Now. I was, all of this is your own network. But a lot of times you have to pay to interconnect with another network, so you have to take that into account. And, as you've heard before, there are congestion costs that you may want to take into account, to make to make the use of your network more efficient and improve the customer experience. And one thing that's often overlooked is billing cost. Lot of times you have a very nice way of pricing a service package, but very difficult to build. And if that's the case, you need a bill, to make money. So when Finally, let's go to the markup. That was the last part of the equation, if you remember. There was price is equal to direct cost, unit cost plus, times a markup. Well, how do you, how do you set a markup? You want to keep it simple so that customers understand. You want it to be consistent across services. So in, in the car industry, for example, the markup is higher on luxury items. Luxury cars as opposed to the, the economy cars. However, that may not be the case in telecommunications. It may turn out that heavy users of the Internet for example, may be more price sensitive than light users and that turns out to be the case. You want to avoid arbitrized price. You're selling a lot of different services to different types of customers. For example, we talked about DSL. It, suppose you're selling a symmetric DSL at four megabits. It's a best effort service, but there's a kind of Cadillac service associated. That's a, which is called DS, DSX services, where you, you provide very high quality of service. Well, yeah, it may turn out if you price DSL so cheap. That others will, move away from the high price. A high quality high price service to this, and you may lose revenue. You wan-, so you want to avoid those arbitrage opportunities. And you want to, you probably want to use markups to reduce congestion. And I think in this course you'll go over quite a bit of state depending on pricing. You want to avoid. In your markup, you want to avoid stranded investment. This is something you won't get to in this class. But let me give you a simple example. Suppose you do some special construction for a wireless carrier. You're building a special line from their tower back to their mobile switching office. Now suppose they say to you, the wireless company says, I only want it on a month to month basis. Give me a price, month to month. Well that tells me, if they wanted to, they could walk away from the service and build their own. Well, I just invested quite a b-. Bit of money in this, in this in this special this construction, if you want to have that flexibility I'm going to raise the price. And you'll find that in option theory, I think there's a course even in the engineering school on that. But that, that's often we often take that into account. And you want to develop a glide path for pushing people from legacy services that are high maintenance. There are still analogue dedicated lines out there. And people still use them, well you try and you can't, you have to maintain them, and the equipment isn't there any more, you have to keep raising the prices to encourage these customers to move to a more efficient type of service. And you want to align your prices with other carriers. You don't want to have your DSL rates sit $200 and everybody else is at $ten. So, you have to take those and, and finally. This is pricing in, in of networks, communication networks, is subject to a lot of regulation, and you have to take that into account. There's you can't run afoul of public policy. There's comparable services, comparable prices, it's part of Telecommunications Act. If you price in such a way that it looks like your not being neutral in the market place, somebody will come after you, and if you're pricing, if your pricing technique requires a lot of deep packet inspection that may be used elsewhere, well there may be some privacy constraints. You have to be careful about that. So let me end up with a specific example, And it illustrates how the, the, the service provided and regulation coming to play. This is the example of a DSL where we have a 25, Uploads, megabit upload speed and 50 download. Now to provide this type of service you need to, have fiber to the home, basically. Not necessarily, but in general. Now look at this, there's a, a link cost, a, a line cost from the customer's premises back to the, let's say the telephone company's office. Now if you notice I have a zero-dollars there. Obviously that cost money, that, that link. Well, it turns out, because of regulation, if you buy. Local service, local telephone service from the local telephone company, and you buy DSL above it, you're not charged for the line for DSL. So your only charged for the electronics. Next we have a port cost. There's, if there are a lot of, of Central office and field equipment that's necessary to provide DSL service. You multiplexing in the field, de-multiplexing and so on. Well the port costs are $649.52 for this particular type of customer. Then there are transport costs. Remember the transport, a lot of, it's, the backbone is a ring and it's transporting service, all sorts of services so it's a combined, a lot of traffic is combined on there. Well, let's say somebody off, Off to the side has figured out what the, what the transport costs are per megabit per mile for this. This, this backbone this backbone infrastructure. Well, then we look at DSL and we find out that the average transport is 9.25 miles. And there's that, there's an over subscription rate of twenty to one. In other words, if you have twenty DSL customers, they're not all using the backbone at the same time. So you can actually have twenty of these types of customers and they only use one equivalent type of cert amount of capacity in the backbone. So we have an over subscription here of twenty customers. With video that's going to come way down, the real time video. But when this was done it was twenty customers. And we get $.03 per mega bit. And do you remember the direct cost factor? There's a return to investors, plus a little bit extra for operating cost and that's 15.9, 49%. per year. So, now we take all of this together. We take, the, the direct investment. That's line one and line two. Times line four divided by twelve, the direct cost of factor, divided by twelve to make it a monthly charge, because the rate of return is per year. And then, we add in the three cents per megabit times 50 megabits download, and that gives you a transport cost per month. And that gives you the direct cost, Unit cost associated with DS, this particular DSL. The actual rate we set was $21.17. So the markup is 109% for this type of customer just based on market conditions.