Monday Digest #110
Weekly summary of finance, economic and tech news. Vision from DreamTeam. With 💜
#macro #finance
WSJ: President Donald Trump’s administration is considering changing the way that the US calculates its trade deficit. The new way of thinking would leave out what are known as re-exports, or exports of goods originally imported from another country, from the exports side of the equation while still counting the good as an import.
The effect of this would be a massive ballooning of the current US trade deficit, according to economists, which would allow the Trump team to paint the US as a loser in the international economy.
This would allow Trump to point to an ever-growing trade imbalance as further proof that trade deals need to be negotiated and that his more protectionist trade stances are warranted.
Bloomberg: Amid trade deficit, which was formed in the U.S. with major trading partners such as China, the deficit with NAFTA partners is relatively small. With most of its 75 trading partners, the US recorded a trade deficit; in 2016 the deficit reached $743 billion with imports exceeding exports 1,5 times. Therefore, the revision of such agreements, like NAFTA, will not bring US to parity of aggregate trade balance.
Simon Lester, analyst at Center for trade policy at Washington’s Cato Institute:
In terms of how it compares to the existing approach, this methodology would not appear to change the overall trade balance. However, it could shift the numbers around a bit between countries. For example, it could make the trade deficit with China lower and the trade deficit with Mexico higher. Of course, most people would say that bilateral trade deficits don’t matter, but the goal here may be to get some new numbers and use them for political purposes, perhaps to help push for a “tougher” trade deal with Mexico. Will this strategy work? The more interesting question for me is what exactly these proposed NAFTA changes will be, as they are probably coming regardless of the trade deficit figure.
US stocks continue to increase: capitalization of the S&P 500 index crossed the mark of $20 trillion (106% of the US GDP), this growth is unsubstantiated as company profits do not grow.
If you look at the net profit, it reached the levels of 2011 (see chart below). Stock index during that time increased by almost 90%. As for the profits, then at the end of last year, the companies increased this figure by only 0.4%, which is significantly below the inflation at 2.8%.
As a result, the ratio of price to forecasted profit of the S&P 500 index is around 21, which is the highest since 2009. The probability of growth of corporate profits is estimated by experts at just 15%.
Robert Naess, who manages 35 billion euros ($37 billion) in stocks at Nordea Bank AB:
“Investors shouldn’t be fooled by top line sales growth as profitability is set to be squeezed by rising wages amid declining unemployment, the fund manager said. With margins already high, corporate earnings estimates will have to come down.”
The fund this year has boosted its stake in EBay Inc. while its biggest increases last year included Walgreens Boots Alliance Inc., Walt Disney Co., Verizon Communications Inc. and Apple Inc.
Vice-President Mike Pence at a press conference (20 Feb.) after a meeting in Brussels with the head of European Council Donald Tusk said that the United States intends to seek new ways for relations with Russia.
Mike Pence:
“We will search in new ways for new common ground with Russia.”
He also said the US would continue to hold Russia accountable over the Ukraine conflict.
Bloomberg: Australia’s government sold A$11 billion ($8.5 billion) of 11-year debt notes in its biggest-ever bond transaction, as investors hungry for higher yields set aside concerns stubborn budget deficits will cost the nation its AAA credit rating.
It’s the third time in less than six months the South Pacific country has set a new borrowing record. It exceeds the A$9.3 billion issued at a sale of December 2021 notes last month and the A$7.6 billion from last October’s debut 30-year deal. The November 2028 securities were priced to yield 3.005 percent.
The government has been ramping up issuance as it struggles to rein in its budget deficit and seeks to finance a debt pile that’s expected to top A$600 billion in 2020.
The 2028 bond has a coupon of 2.75 percent and the transaction was managed by Australia & New Zealand Banking Group Ltd., Commonwealth Bank of Australia, Deutsche Bank AG and Westpac Banking Corp.
Mexico’s central bank is preparing to auction as much as $20 billion in foreign-exchange hedges for the first time to bolster the currency without draining international reserves.
The peso climbed 2 percent to 19.9740 per dollar at 3:40 p.m. in New York, erasing losses, as the intervention is equivalent to selling dollars in futures markets to expand hedges being offered to companies.
Bloomberg: Home values in Monaco reached a record 41,420 euros ($44,100) a square meter last year after rising 15 percent, according to preliminary data compiled by the government’s statistics office.
That’s more than double the price of Manhattan co-ops and condos and almost twice the value of a luxury home in London. Almost half of the 33 new homes purchased in the principality in 2016 were bought for 10 million euros or more, the data shows.
#Commodities
Russia shifted Saudi Arabia from the position of the world leader in crude oil production for the first time since March 2016.
Joint Organization Data Initiative:
- Russia increased oil production in December 2016 by 4% compared to the same period of the previous year — to 10.49 million barrels of oil per day.
- Saudi Arabia has increased production by 3% — to 10.46 million barrels per day.
- In the United States oil production has slightly decreased, from 8.9 million barrels per day in November to 8.8 million in December.
Let us remind, that according to the February monitoring of the IEA, OPEC countries basically comply with obligations to reduce the supply of oil. Freezing led to the first over the last two years significant reduction in OPEC production — by 1 million barrels per day and Russia has reduced deliveries by about 100 thousand barrels per day.
#INTERESTING
Machine learning and neural networks came in the consumer services. Artificial intelligence will soon learn to play with real gamers and treat patients.
- Personal assistant
ABBYY founder David Yang is sure that soon each person will have digital assistants:
“They will be able to quickly find a required file in a mail or phone number of a partner, whom you met at the conference, but can’t remember his name.”
Personal assistant Findo (website) is already able to do this. Findo analyzes public information in the Internet, the contents of the electronic mail, data from notes and cloud files. In the development of Findo ABBYY has invested $3 million and $4 million project has attracted from venture capital and strategic investors, including Flint Capital and Foxit (a leading provider of software for PDF-documents).
In December 2016 Findo already had about 65 000 users.
David Yang:
“Findo understands queries like “Find presentation from someone from London about health insurance that I received a couple of weeks ago”, so far only in English. But soon it will be able to analyze more types of queries and find documents similar in meaning.”
A few years ago, such a recognition of the meaning of the texts and the images seemed impossible. Now more and more application running on the methods of machine learning handle such a task, including deep neural networks, as in the case of Findo.
The neural network — is trainable systems built on the analogy of a human neuronal network. They gave the opportunity to undertake tasks for which it is very difficult to create a concrete algorithm.
They consist of nodes forming the layers that process information.
New information changes the state of the entire system, passing through the layers of neurons. This process is called learning of the neural network.
Algorithm based on neural networks, for example, can analyze a set of texts in any language and automatically group words that are close in meaning, determine the semantic mood of the text, extract specific entities and relationships between them.
Algorithms for machine learning have become popular in 2016, when the business began to use them in applications that are understandable to consumers.
For example, DeepMind company, which was bought by Google for more than $500 million, has reduced the cooling costs of the data center of the Corporation by 40%. Facebook uses machine learning to detect faces in the photo, for analysis of texts and their translations. In September 2016, Google, Facebook, Amazon, IBM and Microsoft have joined forces to create artificial intelligence that will allow them to share data.
Young companies also join a new market.
- Drilling
For example, the startup Nest Lab from Ufa using the machine learning improves the accuracy of selection measures on oil wells. For each well learning algorithms of robot-program Nest determines the relationship between input parameters and final result.
Timur Imaev, marketing director at Nest Lab:
“A separate algorithm is responsible for approving the input data.
Earlier daily measurements did not coincide with the final monthly ones by 15–25% on average in Russia, and with Nest the error falls to 1%.”
The startup has already launched three pilots with oilers in Western Siberia, where the recommendations of the Nest Lab, as Timur Imaev hopes, will be able to give an increase of 30–50%.
- Cucumbers sorting
Machine learning and neural networks are increasingly used in everyday life. Japanese engineer Makoto Koike launched sorting conveyor for cucumbers on the farm of his parents, where the neural network help to classify each vegetable on a photo depending on its length, ripeness and thorness. Koike used an open solution of Google for work with neural networks and achieved 70% accuracy of recognition of cucumbers in the photo, the algorithm was trained on 7000 images.
- Help the whales
National oceanic and atmospheric administration of the USA now monitors 447 North Atlantic whales (that is how much is left on the planet) from satellites. The system was created after the neural networks have helped to identify whales on thousands of photos taken in different oceans at different times of day, in different weather.
Russian Internet giant “Yandex” will start developing its own technology, which is able to train self-driving cars on public roads. The company is looking for developer of unmanned vehicles and developer-researcher of unmanned vehicles.
The technology of “Yandex” will not only be able to manage one car, but in a way to anticipate the maneuvers of surrounding vehicles, adapting to them in advance by resetting the speed or change lanes. The software part will orientate on readings from various sensors, radars and cameras, which are a part of the modern unmanned vehicle.
