About The Authors

Sunday, May 14, 2017

Chinese assembly flow

I'm back from a week in Detroit with students [schedule here] that included factory tours at the Ford Rouge plant in Dearborn, Giffin in Auburn Hills and Continental Structural Plastics in Carey, OH. Both Giffen and CSP were 2017 PACE Award winners. Even though what they saw is a small sample of the range of processes in automotive parts manufacturing and final assembly, they now have a sense of what goes on.

So here is a really nice video that follows a vehicle through the assembly process. During our trip we didn't see welding robots but we did see a paint shop and robots inserting/removing material from a SMC stamping press. Because those were not metal parts we didn't see e-coat in the paint shop of Continental Structural Plastics in Carey, OH. We did however see the stamping of body panels, which is not shown in this video. We did see the roof liner being inserted in the Ford Rouge plant (though given the large number of things being done, students likely don't recall that one particular step). In any case much of this video should now be familiar to my students. Oh, and China is not much different from the U.S. Production in China is driven first and foremost by quality and speed considerations, not by labor costs, so robots do the welding and the heavy lifting.

NYTimes video

Friday, May 5, 2017

LF Participation: Improvement Continues

Mike Smitka, Economics
Washington and Lee University

...how many older workers can be enticed back into the labor market?...

Rather than headline unemployment rate I follow participation rates because so many would-be workers dropped out of the labor market during the Great Recession. In a couple age brackets we've now returned to normal as judged by the pre-recession average, which was roughly flat over the period 2000-2006. The black line is a moving average, which means that it lags as participation increases. The raw average is now at 99%, but bounces around a lot from month to month. I'm thus conservative in how I approach the data. While I do not correct for this (I don't have age-specific data) the share of the labor force working involuntary short hours is down to 3.3% from a Great Recession peak of 6.0%. Some of this improvement thus includes a transition from part-time to full-time work. It's taken 7 years, which means a lot of personal pain and sidetracked careers. But most of the slack in labor markets has now disappeared.

Of course these data are national averages, so there will be a lot of local variation. One caution: participation by older workers at near historic highs. The aging of the baby boomers means there are many people wanting jobs who a generation ago would have been fully retired. How many of them can be enticed back into the labor market? We'll find out over the next couple years. In the meantime, if labor markets really are tightening, then we should start to find employers offering higher wages to attract and retain workers, first on a regional level, then nationally. I however don't follow those data. I need to start reading the Federal Reserve Beige Book, which provides the qualitative impressions of Fed staff in each of its 12 districts.

Tuesday, May 2, 2017

Data Point: Real Yield Curve

Periodically I've posted graphs showing the future bond yields implied by the difference between (say) the yield on a 5 year bond and a 7 year bond. The Treasury also issues TIPS, bonds whose post-inflation yield is guaranteed. Doing so gives suggests after-inflation returns – the yield on a 2-year bond – of 0.65% in 2022. Over the past 10 years, as per the previous post, real bond yields lay below the real growth rate by about 0.65%. So will the ceiling on real growth over the next 5 years be 1.5%?? This of course is consistent with the argument of Robert Gordon that the US is seeing and will continue to see lower growth than experienced in the first four post-WWII decades. (For reference I also include one measure of inflation, the rate of increase in personal consumption expenditures after lopping off items with the highest and lowest price changes. This continues to trend just under 2%.)

Sunday, April 30, 2017

US Deficits and Sustainable Debt

As Congress and the Trump Administration begin to work on their fiscal program, with the potential for tax cuts and expenditure increases, it's important to think about whether our current deficit is sustainable. Now Japan is fast approaching a point where debt issues will overwhelm their financial system. (My senior capstone read a paper by Hoshi & ItoNote that lays forth that case, arguing that the breakdown point will occur by 2027.) The US is not Japan: we have a growing population, less debt, and smaller deficits. Nevertheless at some point we too will need to put our fiscal house in order.

...we don't need to run a surplus, but the current deficit isn't sustainable...

What follows uses a simple (but standard) arithmetic framework to clarify what matters. As long as debt to GDP is stable, we should be OK, because the demand for financial assets grows with the economy. In general institutional investors such as pension funds hold government bonds for good reasons, and that a particular bond has matured doesn't change that. So they want to buy new bonds to replace the old. In other words, at today's level of debt, the Treasury can "roll over" debt, issuing new bonds to replace old. There's not only no need to repay our debt, financial markets would be hard-pressed to find alternative assets if we did so. Indeed, 20 years ago, under the impact of the Clinton administration's budget surpluses, Federal debt was declining rapidly and there was hand-wringing about how financial markets could function if all the debt was repaid. Some of that public worrying was partisan, used by those who wanted to argue that large tax cuts were OK.

Our economy is also growing. So even if the absolute amount of debt continues to rise, potentially debt to GDP will not. Indeed, that's what happened following WWII. By the end of the war debt surpassed GDP, but fell to just over 20% by 1974. This didn't happen because we ran budget surpluses. Quite the contrary, on average we ran small deficits after 1948. But we did grow, enough to outgrow our debt. But today we're running significant deficits and not growing.

Interest rates matter. In the 1950s and 1960s they were relatively low, so the interest the Treasury paid on our debt didn't offset growth. Today we again have low interest rates, but we also have low growth. So we need to ask whether that changes the situation.

Again, what we want to look at is whether debt is stable relative to GDP. That is, if B is the stock of bonds and Y is GDP, is B/Y growing? On its own – assuming bonds are rolled over – the stock grows with accumulated interest: t+1 = Bt(1 + i), where Bt is the stock of bonds at time t and i is the nominal interest rate. Similarly, GDP grows at Yt+1 = Yt(1 + g) where Y is nominal GDP and g is the nominal growth rate. Hence debt to GDP will grow at:

B(1+i)
Y(1+g)

To put this to use, we need three pieces of information: what is the level of debt, B/Y; what is the growth rate g;, and what is interest rate i. That will give us an indication of whether debt is sustainable, and if not, what level of surplus is needed to keep it within bounds.

The first is easy: Federal debt is approximately 100% of GDP, that is, debt to GDP ratio is 1.0 – convenient for arithmetic, as multiplying by 1 is easy. We then need to know the ratio (1 + i)/(1 + g). When i and g are single digits in percentage terms, as in the US, that ratio is approximately 1 + i - g. In other words, with our debt ratio of 1, B/Y will shrink as long as (1 + i - g) is less than 1. The critical issue then is the value of (i - g). If i > g then our debt level will rise, unless we run surpluses. If i < g then we can run (small) deficits indefinitely, as happened during 1949-1974, yet not see our debt level rise.

Now while it might seem that we ought to be able to earn better than the growth rate, this is fundamentally an empirical question. Thanks to the Great Inflation of the 1970s and 1980s nominal interest rates and nominal growth varied wildly. But real growth and real interest rates stay within fairly narrow bounds, except at the depths of our recent Great Recession. The graph below sets forth those data. Excluding the peak around 2009 we find that the average level of (i - g) is about -0.6%. If we include the peak, the average is roughly 0. Now as the graph below indicates, real long term bond yields fell over the past 15 years and are now on the order of 0.8%. Investors, rightly or wrongly, have not built strong growth into bond prices. So to date there's no evidence that the Fed's ongoing normalization of interest rates will raise real interest rates relative to growth. If so, we can run deficits of 0.6% of GDP forever.

To reiterate, we don't need to run a surplus. However, we do need to bring the budget close to balance. Unfortunately, our current deficit is about 3% of GDP. Now that's a vast improvement over the -10% of GDP level at the trough of the Great Recession. Employment growth and profit growth led to stronger income tax receipts, while the improved employment situation led to a drop in "safety net" expenditures. That combination lowered the deficit by a full 7% of GDP. Unfortunately we can't expect further gains, as profits are now high and (un)employment low. There is however downside potential. So we ought to count on the deficit averaging out at -3.5% of GDP, not -3.0%.

...that means we need to "enhance revenue" by 4% of GDP, not cut taxes...

That does not factor in the aging of the baby boomers, who haven't fully retired and whose healthcare expenses will continue to rise until offset by rising boomer mortality. Such retirement-related expenses will likely come to at least 1% of GDP. Hence we need a fiscal adjustment on the order of 4.0%-4.5% of GDP. Congress needs to "enhance revenue," not cut taxes.

Note: Hoshi, Takeo, and Takatoshi Ito. 2014. “Defying Gravity: Can Japanese Sovereign Debt Continue to Increase without a Crisis?” Economic Policy 29(77): 5–44.

Wednesday, April 26, 2017

Tesla: First the Cash, Then the Crash

Tesla recently (briefly) surpassed both Ford and General Motors in market capitalization. My own prognosis for Tesla's longer-run prospects aside, once the Model 3 launches cash flow will improve markedly. Financial markets may interpret that as validation of the high price they've put on Tesla shares. That however is a predictable result of the nature of cash for an automotive assembler, and will have nothing to do with Tesla itself. But it does mean that you shouldn't short Tesla yet.

...you shouldn't short Tesla yet...

Why do I believe this will happen? As our alumnus Bill Cosgrove emphasized in him many visits to Econ 244, in up cycles car companies spin off stupendous amounts of cash. In down cycles they're cash incinerators. Let's examine one source of that (there are others).

Historically dealerships financed the biggest asset in the auto industry, the inventory that sits on their lots. In particular, they pay for cars as they roll off the assembly line. (To be precise, dealerships' banks pay for them via the financing of "floor plan".) Suppliers contributed, too, because they are paid 90 days in arrears. In the early years of the industry they were the biggest cost of production. From the 1920s that role diminished, as Ford and General Motors integrated vertically into making their own engines and frames and bodies. But over the last 25 years car companies systematically spun off their internal parts into standalone operations, Delphi in the case of GM and Visteon in the case of Ford. So paying suppliers in arrears is again qualitatively important. Now when sales increase, for example up 2% in the month, car companies then take in 2% more cash, but pay out nothing more. It's only after 3 months do the payments to suppliers start to rise. That process continues as long as sales rise, irrespective of underlying profitability. A rising tide raises all ... wait, that's not the right image.

Or maybe it is. Because at some point the tide goes out, and with it goes the cash flow. The process reverses. Now car companies are taking in less money, but they're paying out the same amount or – if sales were rising and not just flat – amounts that will continue to increase for 3 months. Worse, it often takes a while to react, as a car company you don't want to slow production because of one odd month. And so the factory pushes metal onto dealership lots. As sales continue to fall, inventory piles up, and at some point the dealers push back, and stop ordering cars. In other words, inventory adjustment amplifies the pace of the downturn. Cars companies burn through cash. They turn from printing presses of profits into incinerators of money.

So when the Model 3 launches, Tesla will report a significant improvement in cash flow. Tesla investors will get good news for a few quarters, as the first 100,000 cars roll off the line and make their way into the garages of the more well-heeled of the enthusiasts who laid down their $1,000 advance deposit. This shift in finances will appear to validate the case for Tesla's high stock price.

But sales can't keep rising forever, particularly as Tesla will only have this one new model, and in a market where consumers are shunning sedans. At that point Tesla will begin hemorrhaging cash. They will have to pay suppliers more and more even as revenue stagnates and then falls. Worse, because they don't have independent dealerships, their debt will also explode as inventories build up. They won't be able to cut output fast enough, because they will have to keep paying suppliers for the higher level of orders prevalent 3 months earlier. They won't be able to cut sales staff, either, if they want to stay in business. Unlike traditional dealerships, their stores won't have service revenue, because most Tesla's will be new. That will be made worse by the nature of their product, as there won't be the oil changes that help service operations generate profits for dealerships even when car sales tank.

...don't short Tesla – yet...

So don't short Tesla – yet. But that time will come, because of the nature of the auto business.

Sunday, March 26, 2017

NPR Marketplace: When it comes to NAFTA and autos, the parts are well traveled

Mike Smitka
Washington and Lee University

On the Friday 24 March NPR Marketplace did an extensive story on automotive parts and NAFTA. For the show, go to When it comes to NAFTA and autos, the parts are well traveled on the Marketplace website.

Scott Tong put together the story, with travel to Ann Arbor. The story that aired featured extended pieces of interview with Dave Andrea of CAR (the Center for Automotive Research) and with Michael Smitka of Washington and Lee University. Scott did a very good job addressing a complex topic.

This is part of a longer Marketplace series NAFTA explained.

For the interview with me Scott was actually sitting in the back seat of the Honda CRV that he discussed with David. Scott skyped me on my computer, I recorded it on my iPhone and then sent the audio files to him. That process worked well, though I had on headphones to lessen feedback which in retrospect probably didn't improve the sound that much, but did make it harder for Scott to piece together the specific questions he was asking as there was no recording of that.

Saturday, March 18, 2017

Oil Prices and Trucks and ICEs: Implications for Tesla's Model 3

As emphasized in Chapter 11 of Smitka & Warrian (up on Amazon Jan 1st!), battery electric vehicles (BEVs) offer many efficiency benefits relative to vehicles powered by an ICE (internal combustion engine). However, this benefit comes with an up-front cost premium that means that, absent subsidies, they fail to provide a value proposition to purchasers. There is no evidence that will change by 2020.

One aspect is that ICEs continue to improve. A quick glance at the Automotive News PACE Awards finalists for 2016-17 shows that 11 of the 34 innovations up for this year's awards improve ICE efficiency. (Winners will be announced in a black-tie event at the Fisher Theater in Detroit on Monday, April 3rd, just before the start of the big annual Society of Automotive Engineers convention. Yours truly will be in attendance.) So while the cost of battery packs continues to fall, and density to increase, they face a moving target. Turbochargers are widespread, start-stop systems are more robust, parasitic losses are being cut as electric motors replace hydraulics, and variable compression systems are starting to launch.

...this is not the time to be launching a new sedan...

Furthermore, there's some indication that the current approaches to Li-ion chemistry are reaching their limits, though the best lab technologies are years from commercial production so incremental improvement will continue apace. Similarly, economies of scale in cell and battery pack are less than expected, evidenced by the construction of multiple small plants. My reading is that costs bottom out around 100,000 vehicles worth of battery packs a year. Now today no single model is at that point, but that day is not far off. Furthermore, battery margins aren't great, so it's not as though car companies pay exorbitant prices.

In other words, manufacturing isn't like internet businesses, where up-front development is a significant part of overall costs and hence every additional user improves profitability. Elon Musk is a victim of the old and mistaken Stalinist belief that bigger is always better. In contrast there are only a handful of "fine chemical" producers, which is where DOE analysis suggests the profits lie. But that's not the end of the business that Tesla is in, nor is it the focus of the battery joint ventures of Nissan and others.

That leaves two markets for electric vehicles: the luxury end, where price does not dominate the purchase decision, much less fuel efficiency. There other is as a "compliance car" made to satisfy one or another regulatory constraint.

ModelAnnouncementDelivery
Model SJune 2008June 2012
Model XFeb 2012Sept 2015
Model 3July 2014? late 2017 ?

Tesla did very well in the $100K end of the market. Rather than supporting the Model S with a full redesign, Musk is instead moving downmarket. Slowly. After all, the company has limited engineering resources, and so can only come out with a new car once in 3 years. (They are promising an earlier date for the Model 3.) Unlike Jaguar, which will use the experienced contract assembler Magna Steyr for the 2018 launch of its I-Pace electric SUV, Tesla insists on going it alone. That eats up capital it could better use to replace its high-end product, and results in slow and late launches, replete with quality glitches.

So what will the market for the Tesla Model 3 be like? The bottom line is, not very good. First, direct subsidies are under attack (though what if anything the current US administration will actually do defies prediction). That is critical for sales of the Model 3, because sales targets transcend the volume the luxury market can generate. Then there are ZEV emission credits and more generally a car's contribution to meeting CAFE restriction. Tesla can't itself use ZEV credits; they need to sell them. And because they sell no fuel-gulping pickup trucks, what happens to CAFE is not directly relevant. Indirectly, though, the launch of the GM Bolt cuts into the ZEV market, and Trump's promised relaxation of CAFE requirements will shift the overall market even further away from the sedan segment. (Using BEA data, in Feb 2017 trucks ran at 11.2 mil SAAR, cars at 6.3 million.) This is not the time to be launching a new sedan, and Trump is promising to make the market less favorable. California is important enough to GM that they need to sell Bolts there to let them sell more high-margin SUVs and pickups. The compliance car market is not relevant for Tesla.

Then there are gas prices. Saudi Arabia would like OPEC to cut output, but is in no position to do so unilaterally. It provides subsidies to keep its population (and ever-larger royal family) happy, and is running a deficit. It desperately needs revenue, but with only a 10% share of the global market – and US tight shale output staying stubbornly high – a cut in output results in a less-than-proportional rise in price. Even with inelastic demand (say, -0.3), revenue falls. So it needs to keep pumping. Nor are US drivers helping, as Bloomberg points out. That makes it even harder to sell a BEV on the basis of their savings on gasoline purchases.

..Tesla investors are set to learn the hard way...

Now I've no doubt that the Model 3 will be a nice car, though even if it does carry a $35,000 list price, you won't be able to find one that isn't loaded. Since on the basis of history there's every reason to believe that initial production volumes will be quite low, we'll be regaled with stories of high demand, cars snapped up as fast as they can be produced. But can they sell 100,000 in 2018? I doubt it, and a mid-sized car in the volume segment won't deliver high margins in a market where everyone wants a truck. Meanwhile the Chevy Bolt will have been out for a full year, the Nissan Leaf updated and BEV subsidies in China pared. Tesla will have to be very generous on the trade-ins they accept to keep sales flowing. They can bury that in their financial statements for a while. Those in the auto industry know how that ends. Tesla investors are set to learn the hard way.

note that Tesla just raised an additional $1 billion in cash

bibliography

Chung, D., Elgqvist, E., & Santhanagopalan, S. (2016). Automotive Lithium-ion Battery (LIB) Cell Manufacturing: Regional Cost Structures and Supply Chain Considerations (No. NREL/PR-6A50-63354 Prepared under Task No. VTP2.6B01). CEMAC (Clean Energy Manufacturing Analysis Center).

Martin, R. (2016). Why We Still Don’t Have Better Batteries. MIT Technology Review, 119(6), 22–22.

Thursday, March 16, 2017

Intel + Mobileye: Strategic Sense, Financial Nonsense

Mike Smitka
Prof of Economics, Washington and Lee University

Cars employ more and more processing power, from chips that handle keyless entry and kick-to-open functions to engine controls and now ADAS (Advanced Driver Assist Systems). Someday there may even be autonomous vehicles, requiring GPUs (graphics processing units) to interpret sensor data and powerful CPUs to run the algorithms. Intel has a very small automotive footprint, having been in and out of several segments over the past 3 decades, at one time making engine control modules for Ford, and buying Wind River, which had projects with BMW and Magnetti Marelli. As the computer and server markets mature needs to develop a wider footprint. At a broad level the Mobileye acquisition could serve as an entrée into the automotive sector, bringing systems capabilities for interpreting vision data that would bring instant access to most global automotive OEMs and major Tier I integrators such as Delphi.

...automakers would welcome a stronger Intel presence...

The automotive end would welcome a stronger Intel presence as well. Over the long haul, no car company wants to be tied to a single provider. Intel rivals are moving. In 2015 NXP acquired Freescale, now Qualcomm is acquiring NXP. That combination will have over $30 billion in revenue. Renasas and Infineon are out there, too, and in individual niches there are players such as NVDIA for GPUs – that ADAS thing – and infotainment systems. Major automotive players would love to have another strong player that they can pull into development projects.

One challenge is execution: Intel has $60 billion in revenue, Mobileye perhaps $300 million. By reputation Intel tends to swallow up small companies, but not digest them and spread their know-how into the larger organization. Selling into the automotive footprint is not easy, with long lead times, complicated and very idiosyncratic test requirements (your cell phone won't work if it's been out in the desert sun, or put through engine compartment vibrations for a month). Mobileye knows how to do that, and could ease Intel's reentry. They also understand the software and data flows in a vehicle system; selling a stand-alone chip is an uninteresting business. Intel needs to add value by bringing to the table a palette of software tools and interface capabilities that let it provide systems-level solutions. Thus it is important that Mobileye not be split up and spit out. The deal has it remaining a subsidiary based as now in Israel, rather than trying to move it (which means its key personnel) to the US. Instead Intel will move people there. Automotive customers will need convincing that this arrangement isn't temporary, and Intel will have to learn how to tap Mobileye's automotive knowledge. That won't be easy.

...the acquisition makes strategic sense...

So the acquisition makes strategic sense, even if there are non-trivial barriers to successful execution.

It does not make financial sense. Rounding so as to make the arithmetic easy, Intel has a P/E of 15. To pay back the $15 billion up-front price, they thus need to generate $1 billion in profits. That won't happen immediately, so we're implicitly doing a net present value calculation. Thus we should ask whether in 2025 the acquisition will be generating $2 billion in profits. (I used a 15% discount rate; assuming Mobileye is making $100 million, this requires that profits increase at an 80% annual rate.) Now maybe that can happen. Both Intel and Mobileye have fat profit margins, so in principle the combination will need to add incremental revenue of $8-$10 billion. But reaching that level will require continuing high levels of up-front investment, call it research. Unlike with making a chip, it will also require continuing R&D, here with an emphasis on the development side of the equation. This isn't like developing a chip, where all the costs lie up front, and the profits flood in once sales commence.

...yet Intel investors should be hearing a warning BZZZ...

Furthermore, the automotive development cycle is long. For all the hoopla about autonomous vehicles, we're looking at a few hundred experimental cars on the road globally in 2020. If that goes well, work will begin on real cars that will launch in the 2025 timeframe. But they will still be high in cost, and the market uncertain. (At the rate Uber and its ilk are losing money, they won't be customers.) Volume, if any, won't come until 2030 and beyond. The holy grail that will make cars a shell for expensive software systems run on expensive hardware will remain out of reach in a timeframe relevant even for patient investors.

In contrast, ADAS is already here. What isn't known is how well received it will be as it moves downmarket. There also remain major problems with the human interface. Drivers can't zone out on the interstate while lane-keeping and adaptive cruise control are engaged, and then just pop back in when there's a warning signal. The systems have to be fail-safe, or more accurately lawyer-proof. Yes, car manufacturers can try to force a driver to keep their hands on the steering wheel. The immediate market is commercial trucking, and in that footprint Intel has already purchased Peloton Technologies (not Peloton, the exercise equipment company!). However, that segment won't generate sufficient revenue to justify the price being paid for Mobileye. In the passenger setment, the rollout of ADAS technologies has been most aggressive in Europe, which is a more technology-friendly in its regulatory and legal system. Vehicle margins in the EU are thin. The big and profitable vehicle markets are NAFTA and China. In Asia, urban congestion and aggressive (random!) driving habits make the ADAS consumer value proposition challenging: when cars move, the warning systems would be producing a constant BZZZ.

Intel investors should be hearing a warning BZZZ, too.

Tuesday, March 14, 2017

Musk’s Facts Are Wrong (Maryann Keller guest post)

Maryann Keller
Principal at Maryann Keller & Associates LLC

Musk’s Facts Are Wrong Regarding the Termination of NUMMI Joint Venture

Published on February 11, 2017 on LinkedIn
cross-posted with permission

In response to Tesla workers at its Fremont, California assembly plant contacting the UAW to possibly seek out representation, Elon Musk responded that the former GM-Toyota NUMMI joint venture, at the same facility Tesla occupies today, was closed because it was a UAW plant and therefore not competitive. In fact, there is nothing further from the truth and his claim is likely due to the oft-prevailing view of the UAW’s role leading to General Motors’ bankruptcy. 

Tuesday, March 7, 2017

Ride Sharing (Maryann Keller guest post)

Guest Post, Maryann Keller, Principal at Maryann Keller & Associates LLC and Ken Elias

David Ruggles speaks regularly with Maryann Keller; we cross-post from LinkedIn with her permission. She is the single most highly regarded auto industry analyst, and has sat on the boards of rental car companies (and was president of priceline.com) so has a good understanding of the economics of ride sharing in one of its "traditional" segments. David took her on her first UBER ride in Washington DC in February 2016. She's since done research on this new "disruptor." They do huge business but lose money doing it. What can be their end game?

Ride-sharing Apps: Low Fares Can’t Last
Published on March 6, 2017

It’s no secret that Uber, Lyft and other ride-sharing apps offer fares less than taxis; exactly how much less depends on metro area, time of day, and ride length. Lower fares, the ease of using the app to summon a car, and cashless payments have built a large user base for the dominant players, Uber and Lyft. But with both companies losing hundreds of millions of dollars a quarter, how soon will investors demand a path to profitability? (TechCrunch.com reports that sources suggest Uber lost three billion dollars last year!)