Archive
Airline Loyalty Miles Have Become Money, not Tokens
I noticed something recently about the many, many airline loyalty miles that my family has accumulated over the years.
Loyalty miles began as a way for airlines to induce brand loyalty in a market that was very fractured post-deregulation (the U.S. airline industry was deregulated in 1978; the American Airlines and United Airlines loyalty programs were created in 1981…although Texas International Airlines is credited with creating the first loyalty program in 1979). In the Old Days, miles worked something like the punch card at the ice cream store, but instead of getting a free scoop of ice cream after ten purchases, it was a free trip after so many segments flown. Because airlines get compensated basically by the number of passenger-miles they create, the loyalty programs were tied to how many miles you flew. Fly more miles, get more miles. But the redemption was fixed: originally, 20,000 miles got you one round-trip domestic coach ticket anywhere the airline flew.
When you get your free scoop of ice cream, it isn’t the scooper’s decision what flavor you get. It’s yours. With ice cream, that’s no big deal; one flavor costs the ice cream parlor about the same amount to deliver to you as another. But with airlines, the problem is somewhat bigger.
Quantitative aside: experienced rates traders may see an echo of the bond-contract structure where it is the seller of the contract who gets to decide which bond to deliver. This optionality is worth something to the seller, and costs something to the buyer, so the bond contract trades at a lower price than it would if there were no delivery options. In this case, it is the buyer who gets to choose what product the seller must deliver (with limitations, of course). So it is very clear that loyalty programs, at least in the traditional structure where the price of the benefit was fixed at 20k or 25k miles, were very valuable to the customer. So did the customer pay more for a fare than he/she otherwise would, to get miles? We may never know.
When the award was “any flight [other than some blackout dates]” and the cost was “20,000 miles”, the strategy was fairly clear. You wanted to wait until you had to buy a high-priced ticket, and buy that ticket with miles instead. In fact, spending the miles on a $400 ticket had a potential opportunity cost because then you wouldn’t be able to spend them on a subsequent ticket that cost $500. So the strategy was to wait, because the option had value. Moreover, inflation worked in your favor as tickets over time rose. There was no realize cost of carry to penalize not spending the miles…so the strategy was to wait. Your loyalty miles were an inflation-linked bond, whose value was linked to airline fares. Actually, an option on an inflation-linked bond…but I digress.
This has changed.
A few years ago, airlines started varying the amount of miles needed to book certain tickets. Tickets on high-load-factor flights started to cost more. In a way, this was not terrible because it meant that some tickets were available at a higher cost, that previously would have been blacked out. So your 25,000-mile award wouldn’t buy the ticket, but you could get it for 50,000. This was successful, and over time what happened is that ever-finer gradations of mile-award-amounts-needed began to show up.
I took an hour this morning and went on United’s website. I priced economy, non-stop, round-trip tickets for EWR-LAX, EWR-ORD, EWR-DFW, EWR-IAD, EWR-BOS, and MIA-SEA(one stop as there were no directs), for March 24-March 26. I collected the price for each departure time. Then I collected the mileage required to buy the ticket in lieu of cash. The chart of this little experiment is below. The x-axis is the miles needed; the y-axis is the dollar cost, and each dot represents one fare pair.
You may notice that the blue dots are arranged in a surprisingly linear way, at least until 32,500 where it seems there is a cap of sorts. In fact, a linear regression line run through the points produces an r-squared of 0.88, and you can get it to 0.95 or so if you use an exponential curve. But the linear line is instructive because the slope of the line indicates that one airline mile on United is worth almost exactly 2.5 cents. As an aside, I didn’t check other loyalty programs but I would be surprised if the slope of American’s line or Delta’s line was meaningfully different.
The red line is where the old 25,000 award would be. If that was still the cost of a ticket, a buyer would not waste it on the tickets to the left of the line and would only use it on those to the right of the line.[1]
So, let’s call a spade a spade: one airline mile on United is 2.5 cents. When airfares go up, your pile of miles becomes less valuable in real terms. Loyalty miles are now indistinguishable from money, in the air travel marketplace.
Here’s the interesting part. Because loyalty miles are now money, the strategy that you the customer should take completely changes. Before, your best strategy was to wait, allow miles to accumulate, and only use them when prices spiked. Now, because miles are money, your best strategy is to spend them as quickly as you can. They don’t earn interest, so they are a wasting asset in real space. It doesn’t matter if you buy one $800 ticket for 32,000 miles, or two $400 tickets for 16,000 miles each. The value is exactly the same.[2] Ergo, they’re money. Not only that, they’re money that can only be spent on airline tickets, and they have a credit component because if the company goes out of business *poof* there go your miles.
Actually, they can be spent on other things, but the optimal way to spend them is probably on airline tickets. I looked at how many miles I would have to exchange to rent various car sizes from Avis in Newark, for two days starting March 24. I added these dots to the chart below.
So the final moral to this story is: don’t rent cars with airline miles!
[1] Class exam question: draw the consumer surplus that the airline reclaimed by changing the pricing structure.
[2] A small caveat to this would be if the current apparent cap at 32,500 for a coach economy class ticket is fixed, because over time more and more tickets would be pricey enough to be capped. However, I think it is unlikely airlines will hold a cap in that way.
The Monetary Policy Revolution in Three Charts
Over the last few years, I’ve pointed out exhaustively how the current operating approach at the Fed towards monetary policy is distinctly different from past tightening cycles. In fact, it is basically a humongous experiment, and if the Fed succeeds in bringing inflation gently back down to target it will be either a monumental accomplishment or, more likely, monumentally lucky. My goal in this blog post is to explain the difference, and illustrate the challenge, in just a few straightforward charts. There are doubtless other people who have a far more complex way of illustrating this, but these charts capture the essence of the dynamic.
Let me start first with the basic ‘free market’ interest rate chart. Here, I am showing the quantity of bank lending on the x-axis, and the ‘price’ of the loan – the interest rate – on the y-axis. If we assume for the moment that inflation is stable (don’t worry, the fact that it isn’t will come into play later) then whether the y-axis is in nominal or real terms is irrelevant. So we have a basic supply and demand chart. Demand for loans slopes downward: as the interest rate declines, borrowers want to borrow more. The supply curve slopes upward: banks want to lend more money as the interest rate increases.
An important realization here is that the supply curve at some point turns vertical. There is some quantity of loans, more than which banks cannot lend. There are two main limits on the quantity of bank lending: the quantity of reserves, since a bank needs to hold reserves against its lending, and the amount of capital. These are both particular to a bank and to the banking sector as a whole, especially reserves because they are easily traded. Anyway, once aggregate lending is high enough that there are no more reserves available for a bank to acquire to support the lending, then the bank (and banks in aggregate) cannot lend any more at any interest rate – at least, in principle, and ignoring the non-bank lenders / loan sharks. We’re talking about the Fed’s actions here and the Fed does not directly control the leverage available to loan sharks.
Now, traditionally when the Fed tightened policy, it did so by reducing the aggregate quantity of reserves in the system. This had the effect of making the supply curve go vertical further to the left than it had. In this chart, the tightening shows as a movement from S to S’. Note that the equilibrium point involves fewer total loans (we moved left on the x axis), which is the intent of the policy: reduce the supply of money (or, in the dynamic case, its growth) by restraining reserves. Purely as a byproduct, and not very important at that, the interest rate rises. How much it rises depends on the shape of the demand curve – how elastic demand for loans is.
As an aside, we are assuming here that the secondary constraint – bank capital – is not binding. That is, if reserves were plentiful, the S curve would go vertical much farther to the right. In the Global Financial Crisis, that is part of what happened and was the reason that vastly increase reserves did not lead to massive inflation, nor to a powerful recovery: banks were capital-constrained, so that the Fed’s addition of more reserves did not help. Banks were lending all that they could, given their capital.
Manipulating the aggregate quantity of reserves was the way the Fed used to conduct monetary policy. No longer. Now, the Fed merely moves interest rates. Let’s see what effect that would have. Let’s assume for now that the interest rate is a hard floor, and that banks cannot lend at less than the floor rate. This isn’t true, but for ease of illustration. If the Fed institutes a higher floor on interest rates then what happens to the quantity of loans?
This looks like we have achieved the same result, more simply! We merely define the quantity of loans we want, pick the interest rate that will generate the demand for those loans, and voila, we can add as many reserves as we want and still get the loan production we need. The arrows in this third chart show the same movements as the arrows in the prior chart. The quantity of loans is really determined entirely by the demand curve – at the prescribed interest rate, there is a demand for “X” loans, and since banks are not reserve-constrained they are able to supply those loans.
However, it’s really important to notice a few things. The prior statement is true if and only if we know what the demand curve looks like, and if the floor is enforced. Then, a given interest rate maps perfectly into Q. But:
- D is not known with precision. And it moves. What is more, it moves for reasons that have nothing to do with interest rates: for example, general expectations about business opportunities or the availability of work.
- Moreover, D is really mapped against real rates, while the Fed is setting nominal rates. So, for a given level of a nominal floor, in real space it bucks up and down based on the expected inflation rate.
- Also, the floor is not a hard floor. At any given interest rate where the floor would be binding, the desire of banks to lend (the location of the S curve) exceeds the demand for loans (by the amount of the ?? segment in the chart above). The short-term interest rate still affects the cost to banks of that lending, but we would still expect competition among lenders. This should manifest in more aggressive lending practices – tighter credit spreads, for example, or non-rate competition such as looser documentary requirements.
In the second chart I showed, the Fed directly controlled the quantity of reserves and therefore loans. So these little problems didn’t manifest.
Now, there is one advantage to setting interest rates rather than setting the available quantity of reserves as a way of reducing lending activity. Only the banking sector is reserve-constrained. If there is an adequate non-bank lending network, then the setting of interest rates to control the demand for loans will affect the non-bank lenders as well while reserve constraint would not. So this is somewhat “fairer” for banks. But this only means that non-bank lenders will also be competing to fill the reduced demand for loans, and the non-bank lending sector is less-vigorously regulated than the banking sector. More-aggressive lending practices from unregulated lenders is not, it seems to me, something we should be encouraging but what do I know? The banks aren’t lobbying me to help level the playing field against the unregulated.
Hopefully this helps illuminate what I have been saying. I think the final chart above would be a lovely final exam question for an economics class, but a bad way to run a central bank. Reality is not so easily charted.
Money and Credit Growth Update
Interested in regular analytical reports? See what we offer at https://store.enduringip.com!
.
It has been a challenging few years to be a monetarist. That isn’t because monetarist predictions have failed, but rather because monetarists have had to spend a lot of time explaining why money velocity has been declining (the answer is: low interest rates) and why “printing money” hasn’t led to runaway inflation (the answer is: inert reserves don’t count, but M2 money growth has been growing between 5%-8% for the last 5 years and that would be too fast for stable prices if velocity was stable).
Money velocity declines when interest rates decline because the demand for real cash balances increases when the opportunity cost of those cash balances is low. That is, if interest rates are at 10%, then you won’t leave cash sitting around idle; it becomes a hot-potato and either gets reinvested in term loans or other assets, or spent. On the other hand if term interest rates are at 0%, then what’s the hurry? The chart below (source: Bloomberg) shows the simple relationship since the early 1990s between 5-year Treasury rates and M2 velocity. This is not a mystery – it has been a critical part of monetarist theory since the 1970s.
You can see that there is a modest conundrum, since interest rates bottomed a couple of years ago but money velocity has continued to sag. I don’t see this as a major mystery; it makes sense to me that there could be some nonlinearities in this relationship near and below the 0% level that we just don’t have enough data to resolve. These nonlinearities have certainly made forecasting more difficult and led generally to forecasts that were modestly too high compared with actual inflation outturns. Again, there’s no mystery about why the forecast misses – the mystery is why money velocity has remained low while interest rates have bounced (we believe economic policy uncertainty has led people to hold somewhat higher real cash balances than they otherwise would, but that’s just a hypothesis). At some point, higher interest rates will snap money velocity back as it gets too ‘expensive’ to leave cash balances sitting around. But this hasn’t happened yet.
Meanwhile, money growth has been slowing. It is still rising faster than 5% per annum, which means that if money velocity was stable and potential GDP growth is 2.5% then we would see the GDP Deflator rising at 2.5%. So money growth is still a bit too fast, unless money velocity is going to decline forever. But it is better at 5% than at 8%, to be sure.
Credit growth has also been slowing, as the chart below (source: Federal Reserve) shows.
Now, regardless of what you read credit growth has essentially no relation to money velocity. Obviously, credit growth has been fairly rapid – as money velocity continued to sag – and is now slowing – as money velocity has continued to sag. It is moderately better connected to M2 growth, so it tends to reinforce the notion that money growth is slowing somewhat, but people who are saying that velocity will continue to slow because banks are slowing loan growth need to explain why rapid growth didn’t lead to velocity acceleration. One-way relationships in economics are pretty rare.
I doubt very seriously that M2 growth is about to drop off a cliff. The Fed’s rate hikes and any balance sheet reduction is not going to affect money supply growth while bank reserves are still “abundant,” to use the Fed’s phrase. Banks are neither capital nor reserve-constrained at the moment, so a decline in credit growth is either coming from the supply side as banks voluntarily reduce loan growth perhaps because credit quality is diminishing, or it is demand side as borrowers are not seeing the growth opportunities that require financing. Money growth is still, and always, something to keep an eye on. But, just as changes in velocity dominated changes in money growth when velocity was falling, velocity changes will dominate changes in money growth when (if?) money velocity starts to rise. As the first chart above shows, velocity when interest rates were “normal” was around 1.8 or higher. I invite you to go to the calculator on the Enduring Investments website and play around using a starting money velocity of 1.43 to see what sort of money supply contraction is required to keep inflation low, if velocity returns to 1.80 over some period of time.
And then, realize that M2 has not declined on a y/y basis as far back as the Fed has records on FRED (about 1960). It seems unlikely to do so now. This leaves few low-inflation exit paths as long as money velocity isn’t permanently dead.
I think the decline in credit growth has implications, but they are mainly implications for growth and not for inflation. Along with the weakness that is starting to be seen in some other areas of the economy (e.g. autos, until the hurricanes caused some “forced replacement”), I think this could be seen as a harbinger of a potential recession in 2018.
