Thứ Năm, 28 tháng 1, 2016

Chinese Financial Scandals

Mike Smitka, Economics, Washington and Lee University

We'll continue to see media coverage of financial scandals in China. For example, the FT Alphaville blog has a "Bezzle Watch" on financial institutions there. This should not surprise us on three levels.

First, under the Stalinist planning system that dominated the urban sector in China there were no banks as we understand the concept. Zero. Similarly in the rural sector communes were expected to fend for themselves – even when that meant privation – so again there was no role for finance, though there were institutions that accepted small individual deposits. Reforms began in the early 1980s that opened up space for "modern" financial institutions to operate, though the legal and institutional foundations weren't put into place until 1994-95. That means that no bank has more than 20 years operating experience. Young institutions that have no experienced staff – and cannot hire from elsewhere because such individuals simply did not exist – have control and monitoring issues. It takes time to set up accounting systems, operating standards, and checks and balances against individual behavior. If global institutions have problems restraining rogue traders, then China's challenges are worse.

...with lower growth, many who operated in the shadows will find their operations cast into the light...

Second, as a young system, it is biased towards large borrowers. This is accentuated by the intermingling of politics and business, as bank executives are typically also party stalwarts, leading to conflicts of interest. But other countries – Japan going into its series of financial reforms in the 1980s – also saw this. For one thing, lending to large borrowers is easier: one small group in a bank can lend to firms tied to electrical manufacturers, Toshiba and Hitachi, another to steel firms, what was in the US context called "National Accounts" (which in 1980 meant Fortune 100 borrowers). The administrative cost of lending was thus very low relative to what would be the case in lending to small firms or consumers, where one banker could only lend a fraction of the amount. Which it was difficult to lend at more than the ceiling "suggested" by government guidance, a bank couldn't offset the higher costs of small business borrowing with higher interest rates. This sort of bias in who can borrow leads those who can tap this flow of funds to borrow a lot, and then use those funds in ways that may lie well outside their core businesses. It can also lead to non-bank players in financial markets, such as manufacturers financing inventory for their customers. That can work out fairly well, helping out for example small retailers while the manufacturer-cum-lender can monitor their sales and inventory levels so as to control credit risk better than banks. But it can easily shade into direct and indirect lending for real estate and playing the stock market.

Third, the Chinese financial system suffers from financial repression. That is typical of developing countries, in which interest rates on deposits are held to low levels, which allows loans to be made at low rates. Even the US, we saw that in the 1970s, when Regulation Q kept banks from matching the rates that were available in money market mutual funds. Other options were available. When I got married, my wife and I pooled our savings – paying a penalty to liquidate time deposits early – and bought a Treasury bond. I carried more cash than I'd ever held before to the offices of the New York Fed to do that. But we also bought an MMMF. We weren't alone, and eventually the outflow of funds from banks (particularly savings banks) force the abolition of Reg Q. Anyway, financial repression creates room for sharp operators. At the same time it provides an incentive for non-bank players to get involved in finance, as per the examples of the previous paragraph.

In a "young" financial system, however, the sorts of players found in the US – my wife and I were early customers of Vanguard – are likely small and poorly known, and have management constraints that prevent them from expanding rapidly so as to service large numbers of customers. That challenge is accentuated by the low level of experience that China's new middle class (and nouveau riche) who need to be educated on what is sound and what is not. Furthermore, financial regulation is similarly "young" with a small and inexperienced staff whom the government, with its wont to pay at levels that lag the private sector, may have a hard time retaining. There's thus lots of space for shadow banking, investment trusts that take in money and use it for ... well, transparency is not the norm. Even when those running such operations intend to be honest, they are inexperienced. Lending to local developers and small businesses is easy, getting money back is harder.

Having worked on Japanese issues for many years, we're sure to be treated to many culturalist explanations, Chinese are intrinsically dishonest and so on. But with the amount of money in the economy, and the immaturity of all those involved – depositors-investors, intermediaries, borrowers – we'll see problems galore. The slowdown of the economy will bring them to the fore, as 10% growth allows all sorts of firms to do well enough, even with large structural shifts that render yesterday's businesses trying to provide goods and services unneeded today. At 6% growth, the unlucky and the incompetent will multiply. And those who were bending the rules and operating in the shadows will find their operations in trouble and in the light. We shouldn't be surprised at what we'll now be able to see.

Hence what we see is a normal transition, not peculiar to China, and not a prelude to collapse.

Thứ Ba, 26 tháng 1, 2016

Award season

Patryk Fournier
The first two months of the calendar year are often referred to as award season by the entertainment industry. Although we don’t compete with the likes of Leonardo DiCaprio and Jennifer Lawrence, we still feel honored with the recent accolades and awards being bestowed upon us.

In the category of Best Backhaul Software or Development Platform for Automakers, the winner is… QNX Software Systems.

Thank you so much to Auto Connected Car News and all the people and companies who voted for us in the Tech CARS Awards. We pride ourselves on offering flexible development platforms that enable automakers to deliver unique, branded experiences. Working with leading-edge automakers and Tier 1 suppliers drives us (pardon the pun) to continue upping our game in advanced platforms for infotainment, digital instrument clusters, advanced driver assistance systems (ADAS), and acoustics — including, of course, the recently announced QNX Platform for ADAS and QNX Acoustics Management Platform.

We would also like to congratulate our fellow award winners, Ford and Harman. Ford won for Overall Best Car Infotainment Software by Automaker for their QNX-powered SYNC 3 connectivity system.

And speaking of Ford, the GSMA Global Mobile Awards recently announced their shortlist of finalists. And we just happen to be a finalist in the category of Best Mobile Innovation for Automotive for our work in Ford SYNC 3.

QNX-powered Ford SYNC 3: Shortlisted for a 2016 Glomo Award. Source: Ford
The Global Mobile Awards, newly rebranded as the Glomo Awards, will take place on February 23 at the Mobile World Congress event in Barcelona, Spain.

Thứ Tư, 13 tháng 1, 2016

The Topology of US Elections: why politicians must lie, er, flip-flop to be elected

Mike Smitka, Economics, Washington & Lee Univ

Under the US primary system for Congress and the White House, politicians must flip-flop in order to be elected. Now candidates may be dishonest in the normal sense of the word. But if a candidate truly wishes to be elected – surely most do! – then they must change their positions during the course of a campaign. That's not healthy for our political system.

Economics helps explain why.

Put every voter on a line from right to left – a typical assumption by those watching campaigns.

Thứ Ba, 12 tháng 1, 2016

“I don’t know where I’m going from here, but I promise it won’t be boring”

Patryk Fournier
The quote is from the now late but great David Bowie and is extremely prophetic when you apply it to autonomous driving. Autonomous driving is very much still uncharted territory. Investments in roadway infrastructures are being made, consumer acceptance is trending positive, and, judging by the news and excitement from CES 2016, the future if anything will not be boring.

CES 2016 stretched into the weekend this year and ICYMI there was a lot of compelling media coverage of QNX and BlackBerry. Here’s a roundup of the most interesting coverage from the weekend:

ARS Technica: QNX demos new acoustic and ADAS technologies
The crew from ARSTechnica filmed a terrific demonstration of the QNX Acoustics Management Platform and the QNX Platform for ADAS. The demonstration highlights the power and versatility of the acoustics platform, including the QNX In-Car Communication module, which allows the driver to effortlessly speak to passengers in the back of the vehicle, over the roar of an engine revving at high speed. The demonstration also showcases how the QNX OS can support augmented reality and heads-up displays:

Huffington Post: CES 2016 Proves The Future Of Driverless Cars Is Promising
Huffington Post highlighted BlackBerry and QNX as key newsmakers for advancements in driverless cars. The article notes QNX’s automotive leadership: “The software is actually installed in 50 per cent of the world’s automotive infotainment systems including Audi, Volkswagen, Ford, GM and Chrysler.”

Crackberry: Inside the QNX Toyota Highlander at CES 2016
The folks at CrackBerry filmed a demonstration of our latest technology concept vehicle, based on a Toyota Highlander. The demo focuses on the QNX In-Car Communication acoustics module, which forms part of the recently launched QNX Acoustics Management Platform:



HERE 360: QNX and HERE bring to life a multi-screen experience in vehicles
A blog post from our ecosystem partner mentions HERE navigation and its use in the Toyota Highlander and Jeep Wrangler technology concept vehicles.

Thứ Năm, 7 tháng 1, 2016

Why is software the key to bringing augmented reality to cars?

Guest post by Alex Leonov, marketing director, Luxoft Automotive.

While self-driving vehicles are gradually becoming a reality, more and more of today’s cars roll out from factories featuring advanced driver assistance systems (ADAS). We are quickly getting used to adaptive cruise control, blind spot monitoring, parking assistance, lane departure warning, and many other features that make driving safer and the driver’s job easier. Data from cameras, sensors, and V2X infrastructure feed into ADAS systems, increasing their accuracy and efficiency. These systems are important steps toward fully autonomous driving, but the ultimate responsibility for decision making still lies with a driver.

The more that cars become connected, the more the average driver can be bombarded by information while driving. “In 500 feet make a right turn.” “You have an incoming call from Christine.” “You have a new message on Facebook.” “You are over the speed limit.” This may not be so big of a distraction under normal conditions. But sometimes, when driving in hectic city traffic or in a snow storm, it is critical to keep eyes on the road, while still receiving essential information. The good news is, the technology is already there to remedy this.

Heads up for HUDs
Keeping the driver’s eyes on the road is a priority, and head-up displays (HUDs) can accomplish just that. They project alerts and navigation prompts right on the windshield. Analysts predict an explosive growth of HUDs with the market reaching close to US$100 billion by 2020. The bulk of HUDs are relatively simple combiners, but more advances in wide-field-of-view HUDs are coming soon.

Projecting alerts and navigation prompts directly on the windshield.
HUDs are perfect for presenting information in a convenient, natural way, and giving the driver a feeling of being in control. But HUDs are only as good as the information they display. That is why it is critical to have solid and reliable data processing and decision-making algorithms, running on a reliable OS, that can prioritize and filter data. The resulting alerts and prompts must be communicated to a driver in a clear, transparent way.

Computer vision, also known as machine vision, is a key to processing the endless flow of data. With its human-like image recognition ability, computer vision processes road scenes, and the system fuses data from multiple sources. Add in a natural representation of processing outcomes in the form of augmented reality, while tracking driver’s pupils, and you have a completely new level of driver’s experience — safe and intuitive.

Next-generation driving experience
At Luxoft, we’ve been working on making this experience a reality. The result is CVNAR, a computer vision and augmented reality solution. CVNAR is a powerful software framework containing mathematical algorithms that process a vast amount of road data in real time to generate intuitive prompts and alerts. CVNAR has built-in algorithms for road and pedestrian detection, vehicle recognition and tracking, lane detection, facade recognition and texture extraction, road sign recognition, and parking space search. It performs relative and absolute positioning and easily integrates with navigation, the map database, sensors, and other data sources. A unique feature of CVNAR is its extrapolation engine for latency avoidance.

Detecting and recognizing road signs, pedestrians, traffic lanes, gas stations, and other objects.
CVNAR works perfectly with LCD displays and smartglasses, but it is ultimately built for HUDs. Data from cameras, sensors, CAN, and navigation maps are fused and processed to create an extendable metadata output that describes all augmented objects. It takes a HUD and an eye-tracking camera to implement CVNAR in a vehicle. CVNAR will track the driver’s gaze and adjust the position of the augmented objects in the driver’s line of sight to make sure they don’t obstruct anything important — all in real time.

Alerting the driver to an empty parking spot.
This is not all that CVNAR can do. New car models come packed with infotainment features that take time to learn and memorize. The CVNAR-based smartphone app can help. It turns your smartphone into an interactive guide. Point your phone camera to your dashboard and use augmented prompts to find out more about a particular car function. It can work under the hood, too.

Era of a software-defined car
A modern car runs on code as much as it runs on gasoline (or a battery-powered electric motor). Today, it takes over 100 million lines of software code to get a premium car going, and the amount of software necessary keeps expanding. At Luxoft, we are excited about the car’s digital future, and we work every day to help bring it about, by developing cutting-edge automotive solutions for leading global vehicle manufacturers.

Offering a wide range of embedded software development and integration services for in-vehicle infotainment and telematics systems, digital instrument clusters, and head-up displays, Luxoft has developed User Experience (UX) and Human Machine Interface (HMI) technology for millions of vehicles on the road today. We push the envelope of technology in such areas as situation-aware HMI, computer vision and augmented reality, while Luxoft’s products, the Populus and Teora UX and HMI design tool chains, power the development of award-winning automotive HMIs and slash time to market.

Software holds the key to the future of cars. It is essential to creating a customized user experience in vehicles. With over-the-air updates, software offers unmatched flexibility and scalability. Finally, it takes safety to the next level with its ability to simulate human-like logic through complex algorithms.

You can view Luxoft’s CVNAR solution running on a QNX-based ADAS demo this week at CES, in the BlackBerry booth: LVCC North Hall, #325.



About Alex
Alex Leonov has been in the automotive and IT industry for over 18 years in various business development and marketing roles. Currently, Alex leads the global marketing efforts of Luxoft Automotive.

In the zone — a visit to the QNX concept garage

Guest post by QNX consultant and software designer Rob Krten.

How often have you heard the expression, “If it were easy to do, everyone would do it”? I’m constantly amazed at the things that QNX does with their concept cars. To me, a car is an inviolate object that must be touched only by the dealer (well, ok, I do top up the windshield wiper fluid and I once changed a battery). I don’t say that because I necessarily like to give the dealer money, but I just don’t want to break anything that’ll cost me more to get fixed properly later.

Pushing the envelope, however, means getting right in there and doing stuff. QNX engineers have done this for their technology concept cars — from replacing the mirrors with LCD screens, to getting right into the dash and rebuilding it, to adding cameras into the antenna fin on the roof. It’s nothing for them to rip out the center console and then look at all the wiring and go, “Huh, ok — so we need to lengthen this wire, add a shim here, move this piece,” and so on. They are fearless.

Redoing the dash of the QNX
reference vehicle.
Sometimes the “getting right in there” is physical; other times, it’s software based — such as making a new application that lives in the infotainment stack or that interfaces with a smartphone. Like a “Dude, where’s my car?” feature — when your Bluetooth phone unpairs with your car, the phone records the current GPS position. Later, when you’re looking for your car, your phone can recall this last stored GPS position — this must be where you left your car. Or even simple aids, such as a radio tuner that detects when you are losing an AM/FM signal and automatically switches to the corresponding digital station, so you can continue listening to your favorite station anywhere you drive.

Curious to see what the future holds, and to actually see some of this work in action, I invited myself down to the “garage” at QNX headquarters. It’s at the far end of the building, next to the cafeteria. The hallway is festooned with posters of previous QNX concept vehicles, highlighting success stories in 3-foot-high glory.

The day I visited, there were half a dozen people in the garage, and two vehicles: a Jeep and a Highlander (otherwise known as the QNX reference vehicle and QNX technology concept vehicle). The garage is a combination of software development lab, hardware development lab, simulation environment, and actual garage (but without the greasy/oily smell). I wanted to get a sense of what drives these people, what they do, and how they do it.

Digital analogs
No, not that kind of digital 
display. Credit: Peter Halasz
The first thing I learned was that there are no real limits. They have the freedom to innovate, without preconceived notions about how things should look. For example, a lead designer on the team (let’s call him Allan, because that’s his name), explained how they look at the controls in the car’s dash display area. In the era of analog, the speedometer had a certain look — it was usually a needle rotating about a central point, where the needle pointed to the speed you were going. In the very early era of digitization, car manufacturers changed this needle to a seven-segment numerical display.

Of course, this was a failure, because the human brain is basically analog; it likes to see nice, continuous changes for processes that are continuous — such as the speed that you’re going. Seven-segment digits change too “randomly”; they require higher-level cognitive functions to parse what the individual lights mean and convert that into digits, and then convert that into a “speed” (and then convert that into “too slow,” or “just right,” or “too fast,” and then, finally, convert that into “apply brake” or “press down on throttle”).

Allan pointed out that changing to a digital display didn’t necessarily mean that they have to slavishly follow the analog “physical” appearance (except do it on an LCD display), but that they were free to experiment with “fill concepts” — digitally controlled analogs to the actual controls. We likened it to the displays in military avionics, where the most important information becomes bigger as it increases in importance. Consider a fighter jet at 20,000 feet — the altitude isn’t nearly as important as it as at 300 feet. Therefore, at 20,000 feet, the part showing the altitude is small, and in a less prominent position than it is when the plane is at 300 feet. The same thing with your speedometer: if you’re doing the speed limit, it’s not as important to show your current speed (you’re most likely flowing with traffic) as it is when you’re 20 over (or under).

In this image from the new QNX technology concept vehicle, the digital instrument cluster is warning that a
forward collision is imminent, and that the driver is exceeding the speed limit by 12 mph. 

You could do the same thing with your fuel range — when you have a full tank, the indicator can be off in a corner somewhere. But as you start to run low, the indicator can get bigger or more prominent, to start nagging you to refuel. By having the displays all be “virtual” on a large LCD screen in the dash, the designers have incredible flexibility to create systems that present relevant information when required, and have it move out of the way when something more important comes along. (Come to think of it, this would be an awesome feature to have on turn-signal indicators — after you’ve kept your blinker on for more than 10 seconds, it would start to get bigger and brighter. Maybe then people would stop driving with their turn indicator permanently on.)

Collision avoided: The V2X command center
Also in the lab was a huge (3 by 5 foot) flat-panel touchscreen, mounted at an angle that’s aggressively unfriendly to coffee cups (probably for that very reason). It’s reminiscent of Star Trek’s main transporter control station, but it’s used to control and display the simulation environment’s V2V (vehicle to vehicle) and V2I (vehicle to infrastructure) data. It acts as a command center to control and reveal the innards of what’s going on in the simulation environment:



When I was there, we ran a vehicle collision avoidance scenario. Two vehicles (the Jeep and the Highlander, of course — they’re tied in to the system) were heading on a collision course (one was southbound and one was eastbound in a grid-style road system). Because they have V2V capabilities, both cars were aware of their impending doom. This showed up nicely on the V2V command center control panel — two cars heading towards each other, little red circles emanating from them indicating the realtime V2V “pings.” Of course, in plenty of time, the Jeep slowed down to avoid the collision (the actual brake lights even went on!). The speed, GPS coordinates, direction, and even what gear each vehicle was in were all shown on the master console. Towards the end of my visit I almost had Allan convinced to do another master control console for the OBDII connector so you could interact with all of the information in each car. What can I say? I like front panels. (I’m a reformed PDP-8 collector.)

The V2X command center, which makes its debut this week at CES, provides a bird’s eye view of several V2X traffic scenarios. In this example, V2X allows a vehicle (the Jeep) to detect that a vehicle up ahead (the Highlander) has braked suddenly, giving the Jeep plenty of time to slow down.

The engineers in the concept garage are “in the zone.” They’re working in an environment that encourages innovation. Watch and see what they produce:




About Rob
Rob is president of Iron Krten Consulting, which provides technical leadership services, from software leadership consulting through to security and embedded software products, development, training and contract services. Rob is also engaged by QNX Software Systems to write marketing and technical documentation. Visit Rob's website.

Video: Paving the way to an autonomous future

Lynn Gayowski
Lynn Gayowski
CES 2016 is now underway, and our kickoff to the year wouldn’t be complete without a behind-the-scenes look at the making of our new technology concept vehicle and updated reference vehicle.

The video below follows the journey of building our vehicles for CES 2016 and highlights the technologies we’re using to speed progress towards automated driving — and the list of tech that QNX covers is impressive! It includes advanced driver assistance systems (ADAS), V2X, and augmented reality, not to mention digital instrument clusters, in-car communication, and infotainment:



QNX Software Systems continues to innovate in automotive, with a vision for the evolution of automated driving and a trusted foundation for building reliable, adaptable systems. At risk of giving away the big finale, I think John Wall, head of QNX, sums up perfectly what QNX is on target for in the automotive industry: “We will dominate the cockpit of the car.” It’s a bold statement but we’re already amassing some imposing stats that back this up:

Thứ Tư, 6 tháng 1, 2016

The simpler, the better: a first look at the new QNX technology concept vehicle

Bringing the KISS principle to the dashboard.

Paul Leroux
“From sensors to smartphones, the car is experiencing a massive influx of new technologies, and automakers must blend these in a way that is simple, helpful, and non-distracting.” That statement comes from a press release we issued a year ago, but it’s as true today as it was then — if not more so. The fact is, the car is undergoing a massive transformation as it becomes more connnected and more automated. And with that transformation comes higher volumes of data and greater system complexity.

But here’s the thing. From the driver’s perspective, this complexity doesn’t matter, nor should it matter. In fact, it can’t matter. Because the driver needs to stay focused on the most important thing: driving. (At least until fully automated driving becomes reality, at which point a nap might be in order!) Consequently, it’s the job of automakers and their suppliers to harness all these technologies in a simple, intuitive way that makes driving easier, safer, and more enjoyable. Specifically, they need to provide the driver with relevant, contextually sensitive information that is easy to consume, without causing distraction.

That is the challenge that the new QNX technology concept vehicle, based on a Toyota Highlander, sets out to explore.

So what are we waiting for? Let’s take a look! (And remember, you can click on any image to magnify it.)

The oh-so-glossy exterior
As with any QNX technology concept vehicle, it’s what’s inside that counts. But to signal that this is no ordinary Highlander, we gave the exterior a luxurious, brushed-metal finish that just screams to have its picture taken. So we obliged:



The integrated display that keeps you focused
When modifying the Highlander, simplicity was the watchword. So instead of equipping the vehicle with both a digital instrument cluster and a head unit, we created a “glass cockpit” that combines the functions of both systems, along with ADAS safety alerts, into one seamless display. Everything is presented directly in front of the driver, where it is easiest to see.

For instance, in the following scenario, the cockpit allows the driver to see several pieces of important information at a glance: a forward-collision warning, an alert that the car is exceeding the local speed limit by 12 mph, and turn-by-turn navigation:



Mind you, the cockpit can display much more information than you see here, including a tachometer, album art, incoming phone calls, and the current radio station. But to keep distraction to a minimum, it displays only the information that the driver currently requires, and no more. Because simplicity.

To further minimize distraction, the cockpit uses voice as the primary way to control the user interface, including control of media, navigation, and phone connectivity. As a result, drivers can access infotainment content while keeping their hands on the wheel and eyes on the road.

Thoughtful touches abound. For instance, the HERE Auto navigation software running in the cockpit interfaces with a HERE Auto Companion App running on a BlackBerry PRIV smartphone. So when the driver steps into the vehicle, navigation route information from the smartphone is transferred automatically to the vehicle, providing a continuous user experience. How cool is that?

Here’s a slightly different view of the cockpit, showing how it can display a photo of your destination — just the thing when you are driving to a location for the first time and would like visual confirmation of what it looks like:



Before I forget, here are some additional tech specs: the cockpit is built on the QNX CAR Platform for Infotainment, uses an interface based on Qt 5.5, integrates iHeartRadio, and runs on a Renesas R-Car H2 system-on-chip.

The acoustics feature that keeps you from shouting
The glass cockpit does a great job of keeping your eyes focused straight ahead. But what’s the use of that if, as a driver, you have to turn your head every time you want to speak to someone in the back seat? If you’ve ever struggled to hold a conversation in a car at highway speeds, especially in a larger vehicle, you know what I’m talking about.

QNX acoustics to the rescue! Earlier today, QNX Software Systems announced the QNX Acoustics Management Platform, a new solution that replaces the traditional piecemeal approach to in-car acoustics with a holistic model that enables faster-time-to-production and lower system costs. The platform comes with several innovative features, including QNX In-Car Communication (ICC) technology, which enhances the voice of the driver and relays it to infotainment loudspeakers in the rear of the car.

Long story short: instead of shouting or having to turn around to be heard, the driver can talk normally while keeping his or her eyes on the road. QNX ICC dynamically adapts to noise conditions and adds enhancement only when needed. Better yet, it allows automakers to leverage their existing handsfree telephony microphones and infotainment loudspeakers.



The reference vehicle that keeps evolving
Before you go, I also want to share some updates to the QNX reference vehicle, which is based on a Jeep Wrangler. Like the Highlander, the Jeep got a slick new exterior for CES 2016:



Since 2012, the Jeep has been our go-to vehicle for showcasing the latest capabilities of the QNX CAR Platform for Infotainment. But for over a year now, it has done double-duty as a concept vehicle, showing how QNX technology can help developers build next-generation instrument clusters and ADAS solutions.

Take, for example, the Jeep’s new instrument cluster, which makes its debut this week at CES. In addition to providing all the information that you’d expect, such as speed and RPM, it displays crosswalk notifications, forward collision warnings, speed limit warnings, and turn-by-turn navigation:



The QNX reference vehicle also includes a full-featured head unit that demonstrates the latest out-of-the-box capabilities of the QNX CAR Platform for Infotainment. For example, in this image, the head unit is displaying HERE Auto navigation:



Other features of the platform include:
  • A voice interface that uses natural language processing, making it easy to launch applications, play music, select radio stations, control volume, use the navigation system, and perform a variety of other tasks.
  • A new, easy-to-navigate UI based on Qt 5.5 that supports a variety of touch gestures, including tap, swipe, pinch, and zoom.
  • QNX acoustics technology that enables clear, easy-to-understand hands-free calls through advanced echo cancellation and noise reduction.
  • Cellular connectivity provided by the QNX Wireless Framework, which simplifies system design by managing the complexities of modem control on behalf of applications.
  • Flexible support for a variety of smartphone integration protocols.

Additional tech specs: The Jeep’s cluster runs on a Qualcomm Snapdragon 602A processor and its user interface was designed by our partner Rightware, using the Rightware Kanzi tool. The head unit, meanwhile, runs on an Intel Atom E3827 processor.

ADAS, augmented reality, V2X, IoT, and more
I have only scratched the surface of what BlackBerry and QNX Software Systems are demonstrating this week at CES 2016. There’s much more to see and experience, including a very cool V2X demonstration, IoT solutions for the automotive and transportation industries, as well as ADAS and augmented reality systems that integrate with the digital clusters described in this post. To learn more, read the press release that QNX issued today and stay tuned to this channel.


QNX announces new platforms for automated driving systems and in-car acoustics

Paul Leroux
Every year, at CES, QNX Software Systems showcases its immense range of solutions for infotainment systems, digital instrument clusters, telematics systems, advanced driving assistance systems (ADAS), and in-car acoustics. This year is no different. Well, actually… let me take that back. Because this year, we are also announcing two new and very important software platforms: one that can speed the development of automated driving systems, and one that can transform how acoustics applications are implemented in the car.

QNX Platform for ADAS
The automotive industry is at an inflection point, with autonomous and semiautonomous vehicles moving from theory to reality. The new QNX Platform for ADAS is designed to help drive this industry transformation. Based on our deep automotive experience and 30-year history in safety-critical systems, the platform can help automotive companies reduce the time and effort of building a full range of ADAS and automated driving applications:
  • from informational ADAS systems that provide a multi-camera, 360° surround view of the vehicle…
  • to sensor fusion systems that combine data from multiple sources such as cameras and radar…
  • to advanced high-performance systems that make control decisions in fully autonomous vehicles



Highlights of the platform include:
  • The QNX OS for Safety, a highly reliable OS pre-certified at all of the automotive safety integrity levels needed for automated driving systems.
  • An OS architecture that can simplify the integration of new sensor technologies and purpose-built ADAS processors.
  • Frameworks and reference implementations to speed the development of multi-camera vision systems and V2X applications (vehicle-to-vehicle and vehicle-to-infrastructure communications).
  • Pre-integrated partner technologies, including systems-on-chip (SoCs), vision algorithms, and V2X modules, to enable faster time-to-market for customers.

This week, at CES 2016, QNX will present several ADAS and V2X demonstrations, including:
  • Demos that show how QNX-based ADAS systems can perform realtime analysis of complex traffic scenarios to enhance driver awareness or enable various levels of automated driving.
  • QNX-based V2X technology that allows cars to “talk” to each other and to traffic infrastructure (e.g. traffic lights) to prevent collisions and improve traffic flow.

To learn more, check out the ADAS platform press release, as well as the press release that provides a full overview of our many CES demos — including, of course, the latest QNX technology concept vehicle!

QNX Acoustics Management Platform
It’s a lesser-known fact, but QNX is a leader in automotive acoustics — its software for handsfree voice communications has shipped in over 40 million automotive systems worldwide. This week, QNX is demonstrating once again why it is a leader in this space, with a new, holistic approach to managing acoustics in the car, the QNX Acoustics Management Platform (AMP):

  • Enables automakers to enhance the audio and acoustic experience for drivers and passengers, while reducing system costs and complexity.
  • Replaces the traditional piecemeal approach to in-car acoustics with a unified model: automakers can now manage all aspects of in-car acoustics efficiently and holistically, for easier integration and tuning, and for faster time-to-production.
  • Reduces hardware costs with a new, low-latency audio architecture that eliminates the need for dedicated digital signal processors or specialized external hardware.
  • Integrates a full suite of acoustics modules, including QNX Acoustics for Voice (for handsfree systems), QNX Acoustics for Engine Sound Enhancement, and the brand new QNX In-Car Communication (ICC).

For anyone who has struggled to hold a conversation in a car at highway speeds, QNX ICC enhances the voice of the driver and relays it to loudspeakers in the back of the vehicle. Instead of shouting or having to turn around to be heard, the driver can talk normally while keeping his or her eyes on the road. QNX will demonstrate ICC this week at CES, in its latest technology concept car, based on a Toyota Highlander.

Read the press release to learn more about QNX AMP.



Thứ Hai, 4 tháng 1, 2016

Ford ports SmartDeviceLink to QNX CAR Platform

QNX joins Ford, Toyota, and other industry leaders to help drive new standard for app integration.

Paul Leroux
For as long as I can remember, QNX Software Systems has been at the forefront of integrating cars and smartphones. Through our flexible OS architecture and large automotive ecosystem, we provide automakers and Tier 1 suppliers with the ultimate choice in connectivity options for smartphones and other smart devices. And now, QNX customers will have even greater choice, with the availability of Ford’s SmartDeviceLink (SDL) technology for the QNX CAR Platform for Infotainment.

If you’ve never heard of SDL, it’s the open source version of Ford AppLink, the software that allows Ford SYNC users to access smartphone apps through voice commands and dashboard controls. Ford donated AppLink to the open source community to create a standard way for consumers to interact with smartphone apps, regardless of which phone they use or vehicle they drive.

SDL is quickly gaining industry advocates, including Toyota, UI Evolution, and, of course QNX. What’s more, companies like PSA, Honda, Subaru, Mazda are evaluating it for use in next-generation vehicles.

Why the interest in SDL? Because it’s a flexible, vendor-neutral standard that can benefit drivers, automakers, and developers alike. With SDL:

  • Drivers can interact with apps by using voice commands, steering-wheel buttons, and other in-car controls, so they can keep their eyes on the road and hands on the wheel.
  • Automakers can deliver a consistent app experience across vehicles, while retaining the flexibility to customize that experience for each vehicle brand or model.
  • Developers can create apps that can work across multiple smart devices and multiple automotive brands — which means they have greater incentive to create automotive apps.

SDL for QNX builds on a history of successful collaborations between Ford and QNX, including the QNX-powered Ford SYNC 3 infotainment system. According to Paul Elsila, CEO of Livio, the Ford subsidiary that maintains the SDL open source project, “With its large market share, QNX can play a key role in driving the adoption of auto industry standards, and we are excited to work with them in building vendor-neutral technology that can simplify the integration of smartphone apps in any brand or type of vehicle.”

SDL works with multiple smartphone platforms. Moreover, it is highly flexible: it can work across a full range of vehicles, from entry-level to premium, and across a wide range of displays. It can even be used in systems without displays — for instance, in systems that use a voice interface.

To learn more about SDL, check out the announcements that Ford, Toyota, and QNX issued this morning.

Thứ Bảy, 2 tháng 1, 2016

China's Auto Industry Meltdown: The Last Shall Be First?

Mike Smitka, economics / Washington and Lee

First to exit, that is. The rush to enter China has led to a market with too many players with too many products and too many assembly plants that are too scattered in geography. The logic is reminiscent of the dot.com era, a combination of optimism unbounded by reality tinged with a belief that, in a market where most consumers are first-time purchasers, buying "clicks" today is essential for future profits. (It's also a predictable consequence of China's policies toward the industry, a topic for other posts.) Most of the new entry and the additions to capacity over the past 10 years took the form of a 50:50 joint venture between a Chinese automotive firm and a global producer. It takes two to tango, and "domestic" players were just as eager to dance as latecomers. But 10% GDP growth and 20% industry volume growth weren't going to continue forever.

...after excess entry, the question is who's first to go?...

Well, the future has arrived. Over the past 15 years, with the 2001 launch of the Buick Sail as a turning point, profits have been shrinking. But they were still high – some analysts claimed that 60% of VW's profits were from its various ventures in China, reflecting a combination of still-fat margins and robust volume. Those heady days are now a thing of the past. While the two market leaders, VW and GM, are hardly likely to go away, not everyone can be in the lead.

Now 50:50 joint ventures are intrinsically unstable. In reality contributions to the venture by the two sides aren't going to be equal; these aren't the sort of startups where the two sides provide little more than cash. Interests diverge. When the Chinese partner is tied to local government, the goals are employment and local tax generation rather than profits, which lack political salience. In addition, the contribution to date was handling local hiring and permits, providing land, and putting in the word at the local branches of banks. The foreign partners contributed knowhow (with varying levels of expat staffing), production rights and kits of imported parts to speed launch. They may have also borrowed under their own name to fund construction, while reinvesting rather than repatriating profits. Other joint ventures are with regional (provincial) governments and "national" State Owned Enterprises, with their own priorities and constraints. In any case – in all cases! – the goals of the partners do not coincide.

Foreign partners are currently limited (at least on paper) to a maximum 50% stake. Their de facto voice has been greater, as they not only pocket 50% of profits but also license fees and the margins on imported parts. (Local partners extract income before profits, too – all know the game. But for them part of the payback has included high-wage jobs and board seats.) Back to the paragraph lede "now": now license fees will be competed away, parts imports are turning into local content [global suppliers have been in China longer than many of their customers], and the foreign partners will be asked to contribute cash. And neither side may have cash, particularly once the joint venture begins burning through it with no end in sight. (Volkswagen and Honda have already cut their investment plans.)

Something will have to give. One is that the 50% ownership stake will be lifted, and global auto firms will be permitted to increase their stake. That option can be exercised only once, and won't remove the problem of negative cash flow. But I wouldn't be surprised if this shows up as part of one or another set of market liberalization measures, bundled into a package of measures that support capital flows and yuan globalization, as the fate of a few joint ventures matters less to Beijing than steps that highlight their ascendance to the global power stage.

Another outcome will be ventures that get shuttered. This has precedent: two of the first three Chinese automotive joint ventures collapsed in the 1990s. The tale of Beijing Jeep, for example, is well and entertainingly told. (I've used American Wheels as a book in my China's Modern Economy class.) This will be an embarrassment to management overseas but may be spun into "willing to make hard choices" by up-and-coming government officials in China. For them the precedents are even deeper: at one time there were 120 vehicle assemblers in China, and cumulative exits likely total about 100 (in the 2000s new entry continued even as firms exited, and there continues to be new entry in the battery electric vehicle segment and of niche producers such as Borgward.

Who will "win" will be determined in part by the strength of dealership networks. Sales are dominated by a new model effect, as they garner the attention of consumers looking for what is still their first car. Consistent with that, dealership profits stem from new car sales, evidenced by BMW's agreement to pay US$820 million and FAW-Toyota to pay US$200 million to improve the profitability of their dealers, with reports that Mercedes and Audi have done likewise. So as the share of repeat purchases (and used car purchases) rises, and as the "park" ages and repairs rise, while the share of vehicles purchased using credit approaches 50%, firms with networks that encourage brand loyalty and can capitalize on service, F&I and used car sales will do better. Those whose dealers opportunistically entered the market while new car profits were high will have a hard time keeping a presence in smaller cities. (China has 200+ cities with populations over 1 million.)

Due to the structure of dealers in their home markets, my hunch is that GM and Ford will do best, witness the rebates already paid by German and Japanese firms. Meanwhile, luck remains a component. Great Wall and Changan, for example, happened to be in the right place at the right time with crossovers, while among joint ventures Ford and GM Wuling are well positioned. (Now everyone is launching vehicles in those segments.) But domestic brands face what Lauren Brandt and Eric Thun label the "battle for the middle" as their start was as producers of low-price vehicles that perforce earn low absolute unit profits and face pressure from the bottom as the used car market develops and from the top as JVs move from down-market from their initial high-end models.

...who wins will be determined by the strength of dealership networks...

Finally, joint ventures that for the first time have excess capacity and produce vehicles up to global content and quality standards are potential exporters. These include Volvo – which is owned by a Chinese firm but functions like a joint venture – Honda and GM. (Exports by purely domestic Chinese exports have flopped; see a CAAM note.) Such exports won't be enough to drive the overall economy, unlike the vision that Chinese policymakers held for the industry in the 1980s and 1990s. At the plant and product level, however, exports can add sufficient incremental volume to put individual operations solidly into the black.

So we will see the past reappearing in 2016, with a few new twists. After excess entry, the question is who's first to go. The last to enter make good candidates.

China surely has too many firms, brands and models. The shakeout will be brutal.

I've not loaded the above with numbers. I have a paper replete with data under submission to a journal that draws on the import substitution industrialization literature to explain the "unusual" structure of China's industry. I'll be updating details as that paper goes through the referee process. In addition, I've the 2015 edition of the yearbook of the Chinese Association of Automobile Manufacturers on order from Amazon China. (My ability to read Chinese is improving rapidly!) In other words, most of my numbers are from non-systematic media sources, which as an economist is not the way I like to put together data.
The orders of magnitude of the data are clear. Domestic capacity is about 30 million units and growing while sales will come in at 22 million units, produced by an industry footprint that includes 34 joint ventures with 18 different partners, and another 20 or so purely domestic ventures. (The number is still rising: Renault will launch their first vehicle only this coming March 2016.) These ventures include ones in remote locations (Urumqi in Xinjiang, in China's far west, and Hainan Island, China's southeasterly extreme) that perforce have high logistics costs.
Sales are fragmented. In CAAM data, production over Jan-Nov 2015 came to 19.5 million units (including small producers gives a Jan-Nov total of 21.8 million), out of which 11.2 million units are passenger cars. The top 10 models in the three main segments – cars, SUVs and MPVs – accounted for 24%, 34% and 82%, respectively. Yet only one model topped 500,000 units, the SAIC-GM-Wuling Hongguang [宏光 "Great Light"] at 579,000 units. At the same time among the top 10 enterprises VW accounted for 3.14 million units and GM for 3.22 million units. No producer accounted for as much as a third of the market. Meanwhile, trying to fight up from the bottom are a host of purely domestic brands that now hold 41.3% of the market. And then there's a wave of electric vehicle ventures that in total are minuscule but are the object of a major policy push including the construction of public charging infrastructure. Now we ought to expect more firms, brands and models in China than in NAFTA or the EU, but basic industrial organization models suggest numbers of firms and brands increase as a square root of market size, not one-for-one. China surely has too many firms, brands and models. The shakeout will be brutal.
  • Brandt, Loren, and Eric Thun (2010). "The Fight for the Middle: Upgrading, Competition, and Industrial Development in China." Working Paper. University of Toronto, Department of Economics.

Thứ Sáu, 1 tháng 1, 2016

Methodology for Calculating Demographics-Corrected Normal Employment Level

reposting from my no-longer-active US and Economics blog, original was November 2012

The Great Recession entailed a huge rise in unemployment; that is easy to track, as it is prominently featured in the monthly Bureau of Labor Statistics releases and is soon thereafter up on the St. Louis Fed FRED data site. Almost as well known is the rise in workers on (involuntary) short hours. That sort of correction is standard, reflected in the "U-6" series of "alternative measures of underutilization."
During the current US recession workers also dropped out of the labor force in unprecedented numbers. While that is a major component of adjustment to business cycles in Japan (and unemployment a smaller component), that has not been the case for the US. During the 2001 recession, employment as a share of the population for the middle of the labor market (ages 30-54) fell by 1.7 points. In contrast, between January 2007 and January 2010 the ratio fell by 5.1 points.
During the past decade, however, the age composition of the population shifted markedly; above all, the baby boomers are now entering retirement. This makes it more difficult to summarize in a single number. But it also turns out that the dynamics across different cohorts are quite different. My own prior was that the Great Recession led to a wave of early retirements, which would show up as a drop in the ration of employment to population. In fact, the ratio rose rather than fell.
This brief note focuses on presenting the data.
It also combines age-specific employment rates with Census Bureau population projections to project future "normal" employment.
Data and Methods
Employment rates by age are available as time series from the Current Population Series on the BLS web site. (See the appendix for the series names.) Specifically, data are available by 5-year brackets (age 16-19 through age 70-74 and then age 75+) on a monthly basis beginning in January 1994. Data are not seasonally adjusted, but visual examination suggests these series have only a modest (and irregular) seasonal component so I have not imposed my own corrections. Because I am interested in the impact of the recession, I construct a (Paasche index) project using a base period (here January 2007) with age-specific brackets and the population in each bracket to project future "normal" employment.
∑ (ratiot x popt)
------------------   x 100   =   employment-population index
∑ (ratiobase x popt)
In particular, given the changing age structure of the population, what number of jobs would be needed to return us to the January 2007 ratio of employment to population? (I have not extended data backwards.)
Now these data should be viewed as providing an additional snapshot. I limited my effort to data readily downloadable from the BLS web site. As a result I make no attempt to examine how employment ratios differ by gender, race and education, nor do I distinguish part-time from full-time labor.1 In addition, I have not compared these data to the more commonly used participation data, which are also available by age bracket. No single number or even modest set of numbers can capture more than a few angles of the dynamics of our labor market.
1. Internal BLS studies such as Sok (2010) utilize more detailed breakdowns for older workers; she also references labor economics studies of shifting trends in the transition to full retirement. Similarly, Rothstein (2012) examines youth (un)employment with detailed breakdowns by gender, race and education.
Age brackets
Visual inspection quickly shows that six age brackets from 25-54 are all very similar in behavior. A quick showed that the difference between high and low fell from 6 percentage points in 1994 to a relatively stable 3.5-4.0 percentage points after early 2001. Almost all of that variation is in fact due to slightly higher employment ratios at ages 30-34 and lower ratios at age 50-54; the variation between age 35-39 and 45-40 is under 1.0 percentage points.
In contrast, younger brackets exhibit a sharp fall. The employment rate among high-school age individuals exhibits a large secular decline; that among the age 20-24 bracket a smaller one. Both drop sharply with the Great Recession.
Older brackets show an opposite trend, a long-term secular rise. Strikingly, these are the only age brackets to exhibit no decline during course of recession. Sok (2010) speculates that this trend may reflect the shift towards defined contribution retirement plans.
Pseudo-cohorts
Pseudo-cohort data – comparing behavior in a bracket with the the next-older bracket 5 years later – can highlight qualitative changes in labor markets.2 In Japanese data there is a pronounced rise in female LF participation starting at 20-24, which is then followed with a lag by a rise in the age 25-29 bracket, and now in the age 30-34 bracket. In other words, in the pseudo-cohort women are increasingly exhibiting "career" behavior rather than dropping out of the labor market with marriage and child-rearing. In Japan there are also smaller systematic changes in older male cohorts. 
The data I use here do not break down employment rates by gender. In the more aggregate US data, visual analysis reveals no patterns as striking as those in Japan. Overall there are modest trend toward declining transition probabilities in the 1999-2001, decreasing from above 1.0 toward 1.0. This probably reflects a tailing off of the rise in female employment rates. There is a modest rise in older brackets. For example, in about 70% of the age 55-59 bracket in 1994 were still working in 1999, when they were age 60-64. By 2005 that had risen to 75%. There is also a 5 percentage point rise going from age 60-64 to age 65-69.  This is a restatement of the rise in employment rates among older workers noted above. However, all series are noisy and there are no clear changes after 2005.
2. A true cohort analysis would track individuals across time. Pseudo-cohorts tracks large averages so may be confounded by trends at the individual level that cancel each other, and by changes at frequencies shorter than the available 5-year increments.
Projections of Future "normal" employment
I previously did a quick and crude constant growth rate projection of "normal" employment based on the rate of employment growth prior to the commencement of the Great Recession in early 2007. However, because of the rapidly shifting age structure of the US population, a simple projection will overstate "needed" employment. Indeed, the 2008 National Population Projection suggests that 2007 is the break point; the aging of the baby boomers began exerting its effect just at the start of the Great Recession.
To estimate "normal" future employment I thus use the January 2007 employment-population ratios as a base and multiply the projected population in each 5-year age bracket by the relevant ratio. The employed population grew at a steady 1.0% during 2001-2006; over the 10 years 1997-2006 it grew an average of 1.3%. However, using the population projections with the January 2007 base, employment growth falls, from 0.9% in 2007 to 0.5% in 2013. As a result, using realistic population projections lowers the anticipated number of employed in 2016 by 5.2 million relative to a simple constant growth projection. The decline would be greater if we limited the analysis to workers aged 25-64, as between 2007 and 2017 the aging of the baby boomers leads to a projected 2.4 million rise in workers age 65+.
The attached graphs provide two simple analyses using this series. One compares actual employment relative to the projected "normal" level of employment. Projecting recent net job creation into the future, this shows that the economy will return to normality in early 2019; a similar projection of employment less those on involuntary short hours provides a similar date for full recovery from the Great Recession. Both imply that in the end this process will have required a full dozen years.
Note that the jobs gap is smaller than the current level of unemployment. That is because even in good times an economy exhibits frictional unemployment, reflecting the normal churn in labor markets as young workers enter the market, and current workers quit to search for better jobs and lose their jobs because of structural shifts in the economy and as firms fire people and search for better workers. Even at the peak of the housing bubble – Oct-Dec 2006 – unemployment never fell below 4.4% or 6.7 million workers. As of October 2012, unemployment was 5.5 million workers higher, while my projection of  employment rates suggests that, reflecting exit from the labor force, the true gap is 9.1 million.
Conclusion
Examining employment rates provides a different picture from other metrics, such as age-specific unemployment rates. In particular, employment rates have not fallen among older workers, in contrast to younger workers. Are employers substituting older (and probably part-time) workers for young (and possibly part-time) workers? or does the human capital of older workers render young workers poor substitutes, so that a tightening of labor markets will not lead directly to an increase in demand for younger workers? These are clearly important topics, but I have not searched EconPapers for recent work on these topics. My primary goal was simply to give a quick overview of the data, and then to use them to 
Bibliography
Rothstein, Donna S. (2012). "Young adult employment during the recent recession." Issues in Labor Statistics, 12(02).
Sok, Emy (2010). "Record unemployment among older workers does not keep them out of the job market." Issues in Labor Statistics, 10(04).
Employment Trend vs Census-adjusted Trend
Employment trend subtracting workers on (involuntary) short hours
Frictional Unemployment
Employment-Population Ratios by Age Bracket: Index
Employment-Population Ratios by Age Bracket: Levels
Pseudo-cohort transition rates
Narrowing of Variation in Employment Rates in Core
Projected Employment
January 2007 Age-specific Employment Rates and 2009 Census Population Projections 
Projected Employment Levels
Increment
Percent growth
2000
 137,282,140 
-
-
2001
 138,781,310 
 1,499,170 
1.09%
2002
 140,188,448 
 1,407,138 
1.01%
2003
 141,539,899 
 1,351,451 
0.96%
2004
 142,919,993 
 1,380,094 
0.98%
2005
 144,324,153 
 1,404,160 
0.98%
2006
 145,746,856 
 1,422,703 
0.99%
2007
 147,002,624 
 1,255,768 
0.86%
2008
 148,192,628 
 1,190,004 
0.81%
2009
 149,389,459 
 1,196,831 
0.81%
2010
 150,560,924 
 1,171,465 
0.78%
2011
 151,681,907 
 1,120,983 
0.74%
2012
 152,585,881 
 903,974 
0.60%
2013
 153,392,625 
 806,744 
0.53%
2014
 154,162,670 
 770,045 
0.50%
2015
 154,873,642 
 710,972 
0.46%
2016
 155,574,275 
 700,633 
0.45%
2017
 156,183,742 
 609,466 
0.39%
2018
 156,731,461 
 547,719 
0.35%
2019
 157,273,686 
 542,225 
0.35%
2020
 157,766,969 
 493,283 
0.31%
2021
 158,338,163 
 571,195 
0.36%
2022
 158,846,493 
 508,330 
0.32%
2023
 159,345,716 
 499,222 
0.31%
2024
 159,852,299 
 506,583 
0.32%
2025
 160,311,745 
 459,446 
0.29%
BLS Data Series
Age Bracket
Series Name
16-19
LNS12300012
20-24
LNS12300036
25-29
LNU02324932
30-34
LNU02324933
35-39
LNU02324934
40-44
LNU02324935
45-49
LNU02324936
50-54
LNU02324937
55-59
LNU02300094
60-64
LNU02300096
65-69
LNU02324938
70-74
LNU02324941
75+
LNU02324942
...mike smitka...