With the advent of the era of consumers, daily consumption has become an investment, changing the traditional concept of consumption, and mass entrepreneurship has entered a new stage!

With the progress of human wisdom and civilization, 2020 is the beginning of a new era. The innovation of production technology represented by "Internet of Things+"has arrived, and a brand-new consumption model and brand-new consumption and sales relationship will change our traditional single trading, payment and sharing methods. In this beautiful era, consumers will get rid of the role of consumers in the past production and consumption links, and then upgrade to consumers.

Then what is a consumer?

A consumer is a business person who manages consumption and consumer groups. The emergence of consumers means that operators have always made money, and the era when consumers spend money will become history; The new era requires us consumers to change the traditional consumption concept, set up the role of integrating resources, sharing products, creating commodity value and benefiting from market economy, find more valuable consumption channels and more confused consumers, pay attention to the role of taking the opinions of consumer groups as the leading factor and participate in the profit distribution of commodities together.

Consumption = investment

Consumers are the economic core of any era. Unfortunately, consumers have never participated in the profit distribution of products. Professor Jesuli Chen, a famous economist, put forward a new consumption concept: Consumers can also become capitalists, also called consumption investment. In other words, the consumer’s purchase behavior is no longer as simple as the simple process of buying and selling, but should be regarded as an investment behavior for manufacturers, and can participate in the distribution of benefits or dividends.

Consumer scenario

In fact, consumer behavior has long existed, but people just ignore it. For example, you bought a beautiful dress, which was cheap and good, and you shared it with your colleagues and friends with unbearable joy. Some of them want to buy it but don’t know where it is, so you tell them or make them buy it. Your behavior is consumer behavior, and your role at this time is consumer behavior. It’s just that the boss who sells clothes doesn’t pay you a profit, or even if he gives it to you, it’s not all the profits of the consumer, because he is a dealer and he has a profit.

Profits of consumers

So, where are the profits of consumers? Let’s put it this way, manufacturers and distributors live on the consumption of millions of consumers in Qian Qian, and the profits of manufacturers are irreplaceable, because the profits of producers are the value of labor and belong to the labor value economy. However, the profits of dealers and publicity can be replaced, consumers replace dealers, and consumers share instead of traditional advertising; These two parts of profits are jointly created by distributors and consumers, which belong to the profit economy of consumers and can be distributed and shared.

summary

Consumer is a brand-new market opportunism, which brings not only products but also more opportunities to the society;

Consumers are dominated by "spending the money that should have been spent and earning the money that could not have been earned", which is a brand-new profit distribution rule;

Consumers are just disseminators of (consumption+sharing) opportunities, not responsible for specific operations, and are the best operators of wealth freedom;

Consumers don’t need, have no employees and don’t need management, and they are a business subject with zero risk;

Consumer is a zero-asset business model;

Consumer is a kind of consumption revolution, which makes consumers participate in the distribution of product profits, makes consumers become consumers, makes the distribution more reasonable and makes the society more harmonious.

Dongguan junfenghuang e-commerce co., ltd. took advantage of the situation to create a brand-new online and offline third-party platform, new concept and new model for consumer sales and entrepreneurial sharing benefits.

Dongguan Junfenghuang E-commerce Co., Ltd. is an innovative online and offline sharing platform for third-party e-commerce that integrates economy sharing, consumption, sales and entrepreneurship. It is based on the technological innovation, model innovation and platform innovation led by China characteristics in the new era proposed by the 19th National Congress of the Communist Party of China, and under the background of mass entrepreneurship, innovation, sharing economy and general trend advocated by the state.

Dongguan Junfenghuang E-commerce Co., Ltd. was successfully established on March 28, 2019 after five years of painstaking research by the founder Mr. Wang Dejun.

One: company philosophy

Break through ideas and be brave in innovation. Keep up with the trend of the new era and the pace of the times. People-oriented, do our best to satisfy people’s yearning for a better life. Wholeheartedly build a diversified, multi-industry, all-round and multi-win innovative e-commerce platform. Everyone can participate in the profit distribution of goods, so that consumers are both creators and beneficiaries. Let consumption become investment, so that consumption can accumulate old-age care. Let the people dare to consume, be willing to consume and be able to consume, let consumption drive employment, and let consumption promote social and economic development. Promote the realization of a well-off society in an all-round way, make the motherland more prosperous and strong, and realize the community of human destiny earlier.

Second, the company model

Build a third-party platform that integrates new models and new concepts online and offline to share benefits with sales, consumption and entrepreneurship, actively explore the concept of consumption capital theory, make full use of the value of consumption surplus (profit), and rationally distribute and share the surplus of consumption, so that consumption can increase in value, consumers’ interests can be maximized, and manufacturers and merchants can circulate goods faster.

1. Consumer consumption value-added

The main innovation of this platform lies in the deep use of the surplus value of consumption. The platform records the surplus value to consumers in the form of integral, which is called added value. The added value is divided into limited value and unlimited value, and consumers can make choices according to their own needs.

A limited value: consumers can exchange the value-added points given by the platform to the platform integral mall for the corresponding gifts they like when they consume, so that the consumption gets extra value and the points will also be.

Corresponding deduction, this is called limited value.

B Unlimited value: Consumers can exchange the value-added points given by the platform to the platform contribution value account when they consume, and enjoy the dividends from the system platform as their contribution to the platform (referred to as contribution value). The contribution value will be released into a shared value (referred to as shared value) every day according to the platform sharing mechanism according to the daily income of the platform, and the shared value will be exchanged for the wallet balance according to the system platform mechanism, and the wallet balance can continue to be consumed on the platform or withdrawn to the bank card. (When the contribution value is released as a shared value, the contribution value will be deducted correspondingly, and the shared value will be increased correspondingly. When the shared value is converted into balance, the shared value will be deducted accordingly). As users spend more and more on the platform, the contribution proportion of user account records will be more and more, and the daily dividends will be more and more. If the sales/consumption of the platform is not limited, the dividends of users will not stop. Consumption is like starting a business, consumption is like an investment in the company, and consumption is like a pension. This fully embodies the infinite value of consumption surplus.

Core values of the platform:

Explore the limited value and unlimited value application of consumption surplus

Co-construction, co-creation, sharing and win-win. Fixed expenditure based on income, and the same source of payment and receipt.

Corporate mission

Push China’s good product and good model to the world.

Let China enterprises go global.

Let users start businesses with zero threshold.

Let the people be willing to spend, be able to spend and dare to spend.

Consumption = investment = providing for the aged, benefiting the country-benefiting the people-benefiting the society.

The responsibilities of entrepreneurs

Dedicate wisdom to society and leave love in the world.

Disclaimer: This article is reprinted by our website, aiming to provide readers with more news information. The contents involved do not constitute investment and consumption suggestions, and are for readers’ reference only.

[Editor: Zhong Jingwen]

How to do economic work in 2019? CCTV Spring Festival Evening gave us these ideas.

  The Spring Festival passes quickly, and I will go to work the day after tomorrow. How to carry out the work in the new year, CCTV Spring Festival Evening in 2019 may provide us with ideas and answers.

  Watching and talking about the Spring Festival Gala is a popular entertainment for the people during the Spring Festival. Looking at it from another angle, the Spring Festival Evening is also an excellent starting point to interpret China’s economic trends. The Southern Economic Think Tank tries to interpret the economic signals released by CCTV Spring Festival Evening from three perspectives.

  Visual angle 1

  Four high-frequency words in the Spring Festival Evening


  High-frequency words in CCTV Spring Festival Evening in 2019. Finishing Chen Yubing

  Through the incomplete statistics of the host words, program lines, lyrics and other party elements,Science and technology, poverty alleviation, revival, express deliveryIt has become a high-frequency word in the CCTV Spring Festival Evening in 2019.

  As the host of the Spring Festival Evening said, "Technology has changed life from extravagant imagination to real daily life". In the Shenzhen sub-venue, the embodiment of "science and technology" is more obvious, from the host words, program arrangement to stage design, which is very scientific and technological. An international science and technology innovation center with global influence is also one of the five strategic orientations of the central government for Guangdong-Hong Kong-Macao Greater Bay Area where Shenzhen is located.

  In the party, it is more symbolic to realize the first 5G network transmission of 4K ultra-HD content in China:2019 is regarded as the first year of 5G commercialization, and the "digital economy" is expected to leverage 5G to achieve rapid development..

  According to the China internet plus Index Report (2018) released by Tencent Research Institute, the national digital economy in 2017 was 26.70 trillion yuan, up 17.24% year-on-year, accounting for 32.28% of GDP.

  Compared with 4G, the characteristics of 5G, such as wide-area coverage, high connection density, high hotspot capacity, low delay and high reliability, can not only meet people’s daily needs for high-speed networks, but also realize the "Internet of Everything", which opens the door for the large-scale popularization of China’s industrial Internet and urges the transformation of old and new kinetic energy in economic development. In other words, 5G empowerment will comprehensively accelerate the digitalization process of China’s economy and society.

  Getting rid of poverty and becoming rich, and building a well-off society in an all-round way is an important task for the central government to read here, and it is of course fully displayed on the stage of the Spring Festival Evening. The "building canals, greenhouses, green ecological parks and food processing plants" mentioned in the Spring Festival Evening sketch are the important contents of industrial poverty alleviation. In 2019, fighting the tough battle against poverty is not only an important economic task, but also an important political task, which not only contains new opportunities and new space for economic development, but also is a powerful answer for China to face the problem of unbalanced and insufficient development.

  There is a familiar prop on the stage of the Spring Festival Evening — — Fuxing train. During this period in Spring Festival travel rush, hundreds of millions of people returned home by rail. The high-speed train is not only a business card made in China, but also the best embodiment of the speed in China.

  It is reported that this year, the national railway investment in fixed assets has maintained a strong scale, ensuring the production of 6800 kilometers of new lines, including 3200 kilometers of high-speed rail.By the end of 2019, the number of Fuxing vehicles put into operation reached 850.. With the improvement of modern transportation network with more high-speed rail and higher efficiency, the logistics cost is expected to be further reduced, and the flow speed of people, logistics, information flow and capital flow is expected to be further accelerated, making it a lubricant and accelerator for China’s economic development.

  Combing this year’s Spring Festival Evening sketch shows will find an interesting phenomenon:Courier has become the hottest profession, and together with cleaning aunts, advertising practitioners, criminal police, house sales agents, and drivers, it forms a professional group that focuses on the Spring Festival Evening.. The general feature of these occupations is that they are all related to urban life and are front-line participants and builders of modern economy.

  Among them, couriers have more distinctive characteristics of the times due to the rapid development of online consumption.

  Just as hundreds of millions of people return home or travel, the same huge logistics is passed to thousands of households through the hands of countless courier brothers. As a new professional group born with the urbanization process, such as courier brother, how to better improve efficiency and their own income and integrate into urban development will be another important topic for China’s economic development.

  Angle of view 2

  Selection of three Spring Festival Gala venues

  The selection of three sub-venues in the Spring Festival Gala is also very interesting.

  From an economic point of view, tea in Jinggangshan, cars in Changchun and robots in Shenzhen respectively represent three industrial forms: traditional agriculture, modern manufacturing and intelligent manufacturing.. Three cities and three kinds of products from the old revolutionary base area in central China, the heavy industry base in northeast China and the southeast coast appeared on the stage of the Spring Festival Evening at the same time, which itself is a microcosm of China’s industrial economy.

  This arrangement supports the basic pattern of the whole Spring Festival Evening in the expression of China’s economic stories. The program arrangement of the three sub-venues has strengthened this point.

  The song and dance "Time No.1" in Changchun FAW Branch of Jilin Province has become an all-round exhibition of China FAW products. The deformed robot modified by FAW cars and the world’s largest indoor remote-controlled aircraft model show the great ambition of the old industrial base to speed up the optimization and upgrading of the economic structure.

  In 2016, the General Secretary of the Supreme Leader visited Shenshan Village, Mao Ping Township, Jinggangshan City, which appeared in the Jinggangshan Sub-venue, and now it has been lifted out of poverty together with six other administrative villages in the township. The Central Economic Work Conference proposed that efforts should be made to increase the supply of high-quality green agricultural products. Jinggang Mountain is green, and it is one of the famous teas in Jiangxi Province. The song and dance "Please Tea Song" is all in full of green, from lyrics to costumes and stage design, full of tea fragrance, which is equivalent to using the super IP of Spring Festival Evening to make a favorite product promotion.

  The song and dance "Fantasy of Youth" in Shenzhen Branch Hall has become a stage for robots and flexible screens, which fully reflects the strong strength and strong enterprising spirit of Shenzhen and even Guangdong-Hong Kong-Macao Greater Bay Area in the fields of electronic information, intelligent manufacturing and other industries.

  Compared with the first two venues, innovation is the most distinctive theme of Shenzhen Sub-venue in terms of stage layout, technical means of signal output and program presentation, which fully embodies the mission of this special zone city in "promoting the deep integration of advanced manufacturing and modern service industries and unswervingly building a manufacturing power".

  The appearance of three sub-venues in succession fully embodies the development characteristics of regional economy and skillfully expounds the different aspects of China’s economic development.

  At the same time, the seamless connection of the three sub-venues has a special symbolic meaning — —The development between different regions and different economic forms is by no means separated from each other..The development of the western region, the overall revitalization of the Northeast, the rise of the central region and the pioneering development of the eastern region are all indispensable and important contents for the country to achieve coordinated regional development. How to implement these national strategies and promote a more adequate and coordinated development of the national economy is a major point of view for China’s economy in 2019.

  Angle of view 3

  Opportunities for the sinking of the consumer market

  Outside the stage, the list of sponsors of the Spring Festival Gala is an excellent material for a glimpse of China’s economic hotspots.

  From Kangbas and Seagull Clocks in the 1980s to Chinese Bicycles, National pits, Wuliangye and other wine brands in the 1990s, Spring Festival Evening advertisements, as a "barometer of China economy", have reflected the economic changes in China for more than 30 years. It has become a true portrayal of the continuous upgrading of the domestic consumer market and the continuous enrichment of people’s living standards.

  In the CCTV Spring Festival Evening in 2019, Tik Tok, Aauto Quicker and Baidu occupied an important position in extending the content sharing of the Spring Festival Evening with their platform advantages. What attracts people’s attention is that Pinduoduo, who was influenced by negative news, became the sponsor of the Spring Festival Gala.

  Behind this round of sponsor replacement, mobile phones have become the absolute protagonist of network use. According to the 42nd Statistical Report on Internet Development in China, as of June 2018, the number of Internet users in China was 802 million, including 788 million mobile phone users, accounting for 98.3% of Internet users from 97.5% in 2017.

  Compared with Tik Tok and Aauto Quicker, the economic significance of Pinduoduo’s appearance on the stage of the Spring Festival Gala is even more extraordinary.

  The Central Economic Work Conference regards "promoting the formation of a strong domestic market" as an important content of economic work in 2019, and proposes to strive to meet the final demand, improve product quality, and accelerate the development of service industries such as education, child care, pension, medical care, culture and tourism.

  Pinduoduo’s becoming a sponsor of the Spring Festival Evening not only shows that the online shopping field is still in fierce competition, but also shows that the consumer market is further enriched and the channels are sinking.

  With the implementation of strategies such as rural revitalization and the adjustment of urban and rural economic structure towards a more coordinated pattern, the consumer markets in first-tier cities, third-and fourth-tier cities and even rural areas are rapidly opening up, and the upward channels of agricultural products are gradually getting through. These factors have provided unprecedented market space for new e-commerce companies like Pinduoduo.

  The emergence of this phenomenon is supported by clear data.

  Previously, the National Bureau of Statistics disclosed when analyzing the economic operation in 2018.In 2018, the per capita consumption expenditure of urban residents in China was 26,112 yuan, a nominal increase of 6.8%, 0.9 percentage points faster than the previous year; The per capita consumption expenditure of rural residents was 12,124 yuan, a nominal increase of 10.7%, 2.6 percentage points faster than the previous year..

  In other words, the rapid growth of rural consumption and the gradual narrowing of the gap between urban and rural consumption have become new features of economic development. At present, the economy is facing downward pressure, and the rapid development of consumption market, especially in non-first-tier towns and rural areas, is even more critical to economic development.

  The rapid development of new e-commerce such as Pinduoduo also means that the online shopping platform has completed the sinking of the "last mile" of the domestic rural terminal market, which means the rapid development of the digital economy in the consumer field. Accompanied by this, it is the further opening and integration of urban and rural consumer markets, and with the steady progress of building a well-off society in an all-round way, a huge market with a scale of 1.4 billion people with more integrity, coordination and comprehensive digitalization is taking shape.

  In fact, regardless of urban and rural areas, there is still huge market space in many consumer areas. The "health care products for the elderly" suspected of selling fakes in "My son is coming" is both a legal issue and an economic issue — — As China enters an aging society, the consumption market of goods and services for the elderly needs to be improved urgently, and the improvement of its quantity and quality is also the key to promoting endogenous economic growth through consumption. People’s pursuit of a better life is also a great opportunity for China’s economy.

  [Write] Wang Biao, a researcher at the Southern Economic Think Tank  

  [Overall Planning] Huang Yinglai

  [Proofreading] Feng Zhijian

  [Product] Nanfang Daily Southern Economic Think Tank

Power Big Data "Portrait" for Economic Operation Helps Zhejiang’s High-quality Development

"In 2022, the electricity consumption of the whole society in Zhejiang fluctuated greatly, and the overall trend was bottoming out. From the perspective of various cities, in 2022, the electricity consumption of the whole society in all regions of the province will increase positively year-on-year … "Recently, the annual electricity data of Zhejiang in 2022 have been released one after another, giving a" portrait "of the province’s economic operation from the macro level, industrial structure and key areas.
Electricity consumption in Zhejiang in 2022. Photo courtesy of Zhejiang Electric Power
Macro level: fluctuating growth bottomed out and rebounded.
Affected by the Shanghai epidemic in April, the high temperature in August and the adjustment of epidemic prevention and control measures in December, in 2022, the electricity consumption of the whole society in Zhejiang fluctuated greatly. "Fortunately, in the most critical fourth quarter, it achieved positive year-on-year growth in three months." The relevant person in charge of Zhejiang Electric Power said that the overall trend was bottoming out.
Electricity is an important "barometer" reflecting economic and social development. In mid-January, 2022, before the Spring Festival, Zhejiang completely stopped production, and the scale of electricity consumption in the whole society declined rapidly. At the end of March, there was a peak of epidemic infection in Shanghai, and Zhejiang Province, as a neighboring province, was also affected by it. In the middle and late June, the government’s steady growth policy took effect and continued to experience extreme high temperature weather, which continued to grow for three consecutive months. The temperature in September was relatively low, the electricity consumption of the whole society decreased year-on-year, and the decline of industrial electricity consumption narrowed; From October to December, the situation of epidemic prevention and control improved, prevention and control measures were optimized and adjusted, and electricity consumption stabilized and rebounded.
From the perspective of local cities, in 2022, the electricity consumption of the whole society in all regions of Zhejiang achieved positive growth year-on-year. Zhoushan leads other cities with a growth rate of 35.47%, contributing 16.77% of the total social electricity consumption growth in Zhejiang Province, which benefits from the new production capacity of Zhoushan Yushan Petrochemical Company.
Electricity consumption of three major industries and residents in Zhejiang in 2022. Photo courtesy of Zhejiang Electric Power
Industrial structure: creating new kinetic energy by "green" new track "number"
The three major industries are progressing steadily, and their development resilience continues to show; New kinetic energy such as manufacturing industry, the core industry of digital economy, continues to play a leading role. Green smart products such as new energy vehicles, charging piles, solar cells and wind turbines have developed rapidly, and the industrial transformation and upgrading trend has continued … In 2022, under the background of shrinking demand, supply shocks and expected weakening epidemic, Zhejiang maintained a sustained and stable economic market in a difficult environment.
Judging from the data of electricity consumption, only 10 of the 31 manufacturing industries in Zhejiang have achieved steady growth in electricity consumption year-on-year. The reason is that the rising prices of global raw materials market, frequent outbreaks of COVID-19 and other factors have led to a huge impact on the industrial chain supply chain in Zhejiang in 2022, and the overall industry, especially the manufacturing industry, is relatively difficult. However, the situation of Zhejiang’s continuous industrial transformation has been continued, and the electricity consumption of green digital industries has achieved rapid growth. Among them, the electricity consumption of photovoltaic equipment and components manufacturing industry, new energy vehicle manufacturing industry, Internet data service industry and charging and replacing service industry all achieved rapid growth.
Internet data service industry is the focus of developing digital economy in Zhejiang, and the electricity consumption of the industry is concentrated in Hangzhou and Huzhou. In 2022, the industrial electricity consumption in Hangzhou and Huzhou increased by 74.18% and 23.69% respectively, accounting for 93.28% of the provincial industrial electricity consumption. With the continuous advancement of Hangzhou-Lake integration, Deqing, as the national new generation artificial intelligence innovation and development experimental zone and the first batch of provincial digital economy innovation and development experimental zones, will be included in the planning of Hangzhou West Science and Technology Corridor, linking Hangzhou Future Science and Technology City, and complementing high-end science and technology resources, which will help Hangzhou and Huzhou jointly become the core area of digital economy development in the province.
The electricity consumption of the new energy vehicle manufacturing industry is concentrated in Jinhua, Hangzhou, Huzhou and Ningbo. In 2022, the industrial electricity consumption in Hangzhou, Huzhou and Ningbo increased by 78.23%, 23.70% and 78.52% respectively. With the rapid development of new energy vehicle manufacturing, supporting new energy battery industry, charging pile manufacturing, auto parts and other industries are also developing rapidly in Wenzhou and Taizhou. In the case that electric vehicles become the general direction of the future automobile industry, the new energy vehicle manufacturing industry is expected to continue its fiery growth.
The growth of residential electricity consumption in Zhejiang in 2022. Photo courtesy of Zhejiang Electric Power
Key areas: ensuring people’s livelihood and using electricity to help common prosperity
In 2022, the growth rate of the three major industries in Zhejiang slowed down, and the electricity consumption of residents stimulated the continuous growth of electricity consumption in the whole society. In 2022, the domestic electricity consumption of urban and rural residents in Zhejiang Province was 90.739 billion kWh, a year-on-year increase of 22.46%, with an increase of 16.642 billion kWh.
In the summer of 2022, Zhejiang experienced "the strongest high temperature in history", and the highest temperature in over 20% counties (cities, districts) broke the historical extreme. Affected by this, the residential electricity consumption in July and August increased by 39.72 and 5.192 billion kWh respectively, up by 41.24% and 57.72% year-on-year, which accounted for 23.87% and 31.20% of the annual residential electricity consumption growth. In addition, under the background of precise epidemic prevention control in Zhejiang in February, 2022 and the optimization and adjustment of epidemic prevention policies in the whole province in December, the length of residents’ stay at home increased significantly during this period, and the electricity consumption of residents increased by 77.92% and 83.56% respectively, which accounted for 23.32% and 24.73% of the annual electricity consumption growth of residents.
The person in charge of Zhejiang Electric Power said that the sharp increase in per capita electricity consumption is not only affected by multiple factors, but also reflects the high level of electrification in Zhejiang people’s lives. "People dare to use electricity and are happy to use it."
In terms of geographical distribution, the per capita electricity consumption level in Zhejiang coastal areas is generally high, while the per capita electricity consumption level in inland areas is low, showing the overall characteristics of "more east and less west". The average household electricity consumption level also shows the distribution characteristics of "less east and more west", and Hangzhou, Jiaxing, Wenzhou, Huzhou and Ningbo have obvious advantages over other cities.
In addition, "common prosperity" is also reflected in power big data. In 2022, the electricity consumption of the construction industry in 26 counties in Zhejiang mountainous area was 1.208 billion kWh, up by 3.82% year-on-year, and the growth rate was higher than the provincial average, achieving a contrarian growth. This is a positive reflection of Zhejiang’s promotion of mountain-sea cooperation, speeding up the completion of transportation infrastructure and new urbanization construction in mountainous counties. These livelihood projects have boosted the growth rate of electricity consumption in the construction industry, and further enhanced people’s livelihood and well-being.
Reporting/feedback

Training the internet is like buying lottery tickets? Interpretation of the lottery hypothesis of the latest development of neural network pruning

Original: the heart of the Synced machine

The heart of the machine is original

Author: Zhu Zihao

Editor: Haojin Yang

Neural network pruning technology can greatly reduce network parameters, reduce storage requirements, and improve the computational performance of reasoning. And at present, the best methods in this field can usually maintain high accuracy. Therefore, the research on sparse architecture generated by pruning is a very important direction. The idea of this topic is to make an in-depth interpretation of the following two papers and explore the best pruning methods today.

Deep neural networks have achieved great success in the field of computer vision, such as AlexNet and VGG. These models have hundreds of millions of parameters at every turn. Traditional CPU can’t do anything about such a huge network, and only GPU with high computing power can train neural networks relatively quickly. For example, AlexNet model, which won the championship in ImageNet competition in 2012, used a 60-million-parameter network with five convolution layers and three fully connected layers. Even if the top K40 at that time was used to train the whole model, it would still take two to three days. The appearance of convolution layer solves the problem of parameter scale of fully connected layer, but after several convolution layers are superimposed, the training overhead of the model is still very large.

Now that there is a more powerful GPU, it is not a problem to calculate a deeper neural network with more parameters. But in fact, not everyone has several cards. For neural networks with more layers and nodes, it is very important to reduce their storage and calculation costs. Moreover, with the popularity of mobile devices and wearable devices, how to make these models well applied in mobile terminals with weak computing power has become an urgent problem. Therefore, more and more researchers began to study neural network model compression.

Synthesizing the existing model compression methods, they are mainly divided into four categories: parameter pruning and sharing, low-rank factorization, transferred/compact convolutional filters, and knowledge distillation [1].

This paper mainly interprets two papers on neural network pruning. The first paper, "The Lotteries Ticket Hypothesis: Finding Sparse, Trainable Neural Networks", was published on ICLR’19 by MIT team. The lottery hypothesis is put forward: the dense and randomly initialized feedforward network contains sub-networks ("winning lottery tickets"), and when trained independently, these sub-networks can achieve the same test accuracy as the original network in similar iterations. This paper won the best paper award. The second article "De Construction Lotteries Tickets: Zeros, Signs, and the Super Mask" is a deep deconstruction of the lottery hypothesis by Uber AI team.

Paper 1: The Lotterity Ticket Hypothesis: Finding Sparse, Trainable Neural Networks

Original link: https://arxiv.org/abs/1803.03635

introduce

Training machine learning model is one of the most expensive aspects in data science. For decades, researchers have proposed hundreds of methods to improve the training process of machine learning model, which are based on an axiomatic assumption that training should cover the whole model. Recently, researchers from MIT published a paper to challenge this hypothesis, and put forward a simpler method to train neural networks by paying attention to sub-networks. MIT researchers gave an easy-to-remember name-"Lottery Ticker Hypothesis".

The training process of machine learning is one of the compromises that data scientists face between theory and reality. Usually, due to the limitation of training cost, the ideal neural network architecture can not be fully realized for specific problems. Generally speaking, the initial training of neural network needs a large number of data sets and expensive calculation cost, and as a result, a huge neural network structure full of complex connections between hidden layers is obtained. This structure often needs to be optimized by removing some connections to adjust the size of the model. A question that has puzzled researchers for decades is whether we really need such a huge neural network structure. Obviously, if we connect every neuron in the network, we can solve a specific problem, but we may be forced to stop because of the high cost. Can’t we start with a smaller and leaner network? This is the essence of the lottery hypothesis.

Taking gambling as an analogy, training the machine learning model is equivalent to obtaining the winning lottery ticket by buying every possible lottery ticket. But if we know what the winning lottery looks like, can we choose the lottery more intelligently? In the machine learning model, the huge neural network structure obtained in the training process is equivalent to a big bag of lottery tickets. After the initial training, the model needs to be optimized, such as pruning and deleting unnecessary weights in the network, thus reducing the size of the model without sacrificing performance. This is equivalent to finding the winning lottery ticket in the bag and then throwing away the remaining lottery tickets. Usually, the network structure after pruning is about 90% smaller than the original one. Then the question is, if the network structure can be reduced, why not train this smaller network from the beginning in order to improve the training efficiency? However, many experiments have proved that if the pruned network is trained from scratch, the accuracy is much lower than that of the original network.

The idea behind MIT’s lottery hypothesis is that a large neural network contains a small sub-network, and if it is trained from the beginning, it will get similar accuracy to the original network.

Lottery hypothesis

The formal definition of the lottery hypothesis in this paper is that a randomly initialized dense neural network contains an initialized sub-network, which can achieve the same test accuracy as the original network after the same number of iterations at most when training alone.

We regard all the parameters of a complex network as the prize pool, and the subnet corresponding to the above set of sub-parameters is the winning lottery.

More formally, consider a dense feedforward neural network f(x; θ), where the initialization parameter θ=θ_0~D_θ. When descending with random gradient on the training set, F can achieve loss L and accuracy A after J iterations. In addition, consider applying a 01 mask m ∈ {0,1 }| θ| to the parameter θ, and train f(x; M⊙θ), f reaches loss L’ and accuracy A’ after j’ iterations. The lottery hypothesis indicates the existence of m, which makes j < = j (faster training time), a > = a (higher accuracy), ||| m || _ 0 < | θ| (fewer parameters).

How to find the winning lottery ticket

If the lottery hypothesis is correct, then the next question is how to design a strategy to find the winning lottery. The author puts forward a method to find the winning lottery ticket by iteration:

1. Randomly initialize a complex neural network.

2. train this network j times until it converges.

3. Cut off some weight parameters

4. Initialize the remaining sub-networks with the weights in step 1 to create winning lottery tickets.

5. In order to evaluate whether the sub-network obtained in step 4 is a winning lottery ticket, train the sub-network and compare the accuracy.

The above process can be carried out once or many times. When there is only one pruning, the network is trained once, and the weight of p% is cut off. In this paper, pruning is carried out N times iteratively, and the weight of P (1/n)% is cut off each time.

experimental analysis

The author has done a lot of experiments on the fully connected neural network for MNIST and the convolutional neural network for CIFAR10 respectively. Take MNIST experiment as an example here:

Pm represents how many parameters are left in the network. It can be observed from Figure 3 that the performance of sub-networks with different pruning rates is different. When Pm>21.2%, the smaller the Pm, that is, the more pruning parameters, the higher the accuracy. When Pm<21.1%, the smaller the Pm, the lower the accuracy. The winning lottery ticket converges faster than the original network, and has higher accuracy and generalization ability.

It can be observed from Figure 4 that iterative pruning can find the winning lottery ticket faster than oneshot pruning, and it can still achieve high accuracy in the case of small sub-network. In order to measure the importance of initialization in winning lottery tickets, the author keeps the structure of winning lottery tickets and then uses random initialization to retrain. Different from the winning lottery, the re-initialized network learning speed is slower than the original network, and the test accuracy will be lost after a small amount of pruning.

summary

In this paper, the author puts forward the lottery hypothesis and gives a method to find the winning lottery ticket. A subnet can be found by iterative unstructured pruning, and it can be trained faster with the initialization parameters of the original network, but the same performance can not be achieved if the random initialization method is used.

The author also points out some problems in this work. For example, the calculation of iterative pruning is too large, and a network needs to be trained for 15 or more times in a row. In the future, we can explore more efficient ways to find winning lottery tickets.

Paper 2: de construction lotty tickets: Zeros, signs, and the super mask.

Original link: https://arxiv.org/abs/1905.01067

Review of lottery hypothesis

Frankle and Carbin put forward a model pruning method in the lottery hypothesis (LT) paper: after training the network, reset all the weights less than a certain threshold to 0 (pruning), then reset the remaining weights to the initial weights of the original network, and finally retrain the network. Based on this method, two interesting results are obtained.

On the one hand, the performance of a heavily pruned network (with 85%-95% weight removed) has not obviously decreased compared with the original network, and if only 50%-90% weight is removed, the network performance will often be higher than the original network. On the other hand, for the trained ordinary network, if the weights are re-randomly initialized and then trained, the results are equivalent to those before. However, the network with lottery hypothesis does not have this feature. Only when the network uses the same initialization weight as the original network can it be well trained, and if it is reinitialized, the result will be worse. The specific combination of pruning mask (if the deletion right is reset to 0, otherwise it is 1) and weight constitutes the winning lottery ticket.

existing problems

Although the lottery hypothesis is proved to be effective in the last paper, many potential mechanisms have not been well understood. For example, how does LT network make them show better performance? Why is the mask and the initial weight set so tightly coupled that reinitializing the network will reduce its trainability? Why simply choosing a large weight constitutes an effective criterion for selecting a mask? Will other criteria for selecting masks also work? This paper puts forward the explanation of these mechanisms, reveals the special modes of these sub-networks, introduces a variant that competes with the lottery algorithm, and obtains an unexpected derivative: supermask.

Mask criterion

The author sets the mask value of each weight as the function M(w_i,w_f) of the initial weight and the trained weight, which can be visualized as a set of decision boundaries in two-dimensional space, as shown in Figure 1. Different masking standards can be considered as dividing a two-dimensional (wi = initial weight, wf = final weight) space into regions with a mask value of 1 vs 0.

As shown in the figure, the masking criterion is identified by two horizontal lines, which divide the whole area into mask =1 (blue) area and mask =0 (gray) area, corresponding to the masking criterion used in the last paper: keep the final large weight and cut off the weight close to zero. The author calls this large _ final mask, m (w _ i, w _ f) = | w _ f |. The author also put forward eight other mask criteria, and the corresponding formulas are shown in the figure below, which keep the weight of the colored part of the ellipse and cut off the weight of the gray part.

The author has done a series of comparative experiments on these mask criteria, and the results for fully connected and Conv4 networks are shown in the following figure. It can be found that magnitude increase is comparable to large_final, and it will perform better in Conv4 network.

Taking the random mask as the baseline, we can find that those criteria that tend to retain the weight with a large final value can better find the subnet, but the effect of retaining the small weight is poor.

Importance of sign

Now we have explored which weights are better for impairment. The next question is how to reset the remaining weights. The author mainly wants to study an interesting result in the last paper. It works well when it is reset to the initial value of the original network, but it will get worse when it is initialized randomly. Why does reinitialization get worse and what are the most important initialization conditions? In order to find the answer to the question, the author has done a series of initialization experiments.

Reinit: Initialize the reserved weights based on the original initialization distribution.

Refresh: Initialize based on the original distribution of reserved weights.

Constant: Set the reserved weight to a positive or negative constant, that is, the standard deviation of the original initial value of each layer.

It can be found that preserving the initial value of the weight is not as important as preserving the sign. If other initialization methods are used, but the sign is ignored, the effect is very poor, which is similar to random initialization (dotted line in the figure). However, if the original weights keep the same sign, the initialization effect of the three methods is almost the same as that of LT network (solid line in the figure). As long as the signs are consistent, even if the remaining weights are set to be constant, it will not affect the performance of the network.

Supermask

At the beginning, the concept of super mask is mentioned, which is a binary mask. When it is used in randomly initialized networks, it can get higher accuracy even without retraining. Here’s how to find the best super mask.

Based on the above insight into the importance of the initial symbol and the consideration of making the weights closer to the final value, the author introduces a new mask criterion, selects larger weights, and these weights also keep the same sign after training, which the author calls large_final, same sign. And use large _ final and diffsign as control, and the difference between them is shown in the following figure.

By using this mask criterion, 80% of the test accuracy can be achieved on MNIST, while the accuracy of the large_final method in the last article is only 30% at the best pruning rate (note that this is without retraining).

summary

This article gives an in-depth explanation of the last article. By comparing different mask criteria and initialization schemes, we can answer why the lottery hypothesis can perform well. And interestingly, a new "super mask" is proposed, through which high accuracy can be obtained without retraining the subnet. This provides us with a new neural network compression method, which can reconstruct the weight of the network only by saving the mask and random number seeds.

references

1. Cheng, Yu, et al. "A survey of model compression and acceleration for deep neural networks." arXiv preprint arXiv:1710.09282 (2017).

2. Frankle, Jonathan, and Michael Carbin. "The lottery ticket hypothesis: Finding sparse, trainable neural networks." ICLR (2019).

3. Zhou, Hattie, et al. "Deconstructing lottery tickets: Zeros, signs, and the supermask." arXiv preprint arXiv:1905.01067(2019).

4. https://eng.uber.com/deconstructing-lottery-tickets/

5. https://towardsdatascience.com/how-the-lottery-ticket-hypothesis-is-challenging-everything-we-knew-about-training-neural-networks-e56da4b0da27

Author introduction: Zhu Zihao, currently a graduate student of Institute of Information Engineering, Chinese Academy of Sciences, whose main research interests are graph neural network, multimodal machine learning, visual dialogue and so on. I like scientific research and sharing. I hope to learn and communicate with you through the heart of the machine.

This article is original for the heart of the machine, please contact Ben WeChat official account for authorization.

✄————————————————

Join the heart of the machine (full-time reporter/intern): hr@jiqizhixin.com

Contribute or seek reports: content@jiqizhixin.com.

Read the original text

The opening of the TV market in the international film and television market conveys an important "signal"

  (Reporter Li Junna) As an important exhibition and trading link of the Shanghai International Film and Television Festival, the TV market of the international film and television market officially opened in Shanghai Exhibition Center yesterday. Yesterday morning, Zhang Hongsen, deputy director of the State Administration of Radio and Television, and his party visited the pavilion, including the "Belt and Road" theme pavilion, Shanghai Radio, Film and Television Production Industry Association, Xinli Media and other exhibition areas.

  This year, the number of booths in the TV market has increased by more than 30% compared with last year, and the "Belt and Road" theme pavilion has been set up for the first time, and various interactive ways such as national TV program promotion and international buyer exchange meetings have been adopted to provide a multi-platform for exchanges and cooperation between TV institutions in various countries. As in previous years, hundreds of domestic and foreign film and television production companies, publishers and broadcasters set up exhibitions to "drink". As one of the weathervanes of TV dramas in the second half of the year and even next year, dazzling TV dramas also convey the "signal" of the current TV drama market.

  Realistic themes become mainstream.

  In recent years, realistic works closely following the pulse of the times have become the mainstream of creation, which will be further strengthened in future TV drama creation. In this year’s international film and television market, many film and television institutions put up posters of realistic TV dramas on their booths.

  "This year, we will have a number of realistic works." Wang Qiao, vice president of Xinli Media, introduced that Sesame Hutong, which is currently being filmed, expresses the life of a group of ordinary people in Sesame Hutong in the great era. Previously, Xinli Media adapted Yi Shu’s works and filmed My First Half Life, which became a phenomenal hot drama. Next, Yi Shu’s other work "last romance" will be adapted and filmed.

  The industry believes that the popularity of film and television companies robbing the head IP in the past few years is gradually declining, and today’s TV market is returning to the realistic theme of grounding gas. Yaoke Media, which has produced a series of realistic dramas, such as Mind, Divorce Lawyer, Women are not strong and Heaven forbid, Long time no see, has always taken realistic dramas as its main development direction. During the Shanghai TV Festival, Yaoke Media also released realistic dramas such as People’s Property, People Selling Houses, I heard that you like me, Love Like Stars and so on. Among them, "People’s Property" focuses on state-owned enterprises, which was co-written by the famous screenwriter Zhou Meisen and director Shen Yan.

  In realistic drama, the theme of love and marriage is still a new force. The urban love and marriage drama "We Are All Good" produced by Mango TV and Joy Film Media Co., Ltd. was unveiled at the Shanghai TV Festival. The drama focused on the love and marriage topics of the post-80 s and post-90 s, showing the course of the protagonist’s initiative to grow up from life difficulties. Liu Xuesong, the director, has directed The Times of the Storm, The Emergency Doctor and other works, and is good at telling the real theme.

  Online drama has gradually become a climate.

  The trend of online drama surpassing TV series to become the main battlefield of the industry is becoming more and more obvious. From the perspective of the international film and television market, not only is its quantity experiencing explosive growth, but its quality is expected to achieve a qualitative leap.

  In the Shanghai New Culture Media Exhibition Area, the posters of "Demon Journey to the West" and "Mermaid" attract people’s attention, and these two works are inextricably linked with the famous director Stephen Chow. "Among them," Journey to the West "is mainly an online drama aimed at online audiences, and" Mermaid "is positioned as a TV series linked by network and Taiwan." Gu Chen, director of corporate propaganda department of Shanghai New Culture Media Group Co., Ltd., told the reporter: "Although the copyright of Stephen Chow’s works has been bought for adaptation, it will re-hatch and break through in content. The audience will only pay for good content, and the shoddy online drama cannot keep the audience. As a production company, only the best production level can be produced. "

  Interestingly, among these people in the industry, the former "network linkage" is subtly becoming "network linkage". Gu Chen said: "Although the purchasing power of the traditional TV platform still exists, the resource advantage of the network platform is being further strengthened."

  This is also reflected in the film and television incubation project of online literature. Zhao Chen, general manager of Zongheng Literature, told the reporter: "Zongheng Literature, which was hatched the year before last, ranked first in Youku’s online ranking. This incident shows that the value of IP incubation must exist, the platform resources are limited, and some particularly good content may not be displayed in traditional TV channels, so it needs to be derived in multiple dimensions. "

The patent industry is moving from winning by quantity to improving by quality.

In 2017, China’s patent industry rose in quantity and quality, patent creation made steady progress, and China patents began to enter a new era of development.
From quantity winning to quality improving.

Panorama of FAST project. Xinhua News Agency reporter Liu Xushe

In 2017, the top ten enterprises (excluding Hong Kong, Macao and Taiwan) granted invention patents in China: pieces

In 2017, provinces (autonomous regions and municipalities) with more than 1,000 PCT international patent applications: 10,000.

In 2017, the top ten provinces (autonomous regions and municipalities) with invention patents per 10,000 population in China: pieces

Although Chinese enterprises are growing rapidly in the fields of mobile communication, electronic information technology and chemical industry, the number of foreign invention patents in China is higher than that in China in five fields, such as optics, medical technology, engine, audio-visual technology and transportation.

Judging from the number of invention patents that have been maintained for more than 10 years, the number in China is still less than that in foreign countries in 29 technical fields, and it is still necessary to continue to cultivate high-value core patents in the above fields.

Recently, China National Intellectual Property Administration released the information about China’s invention patents and the ranking of enterprises in 2017. The data shows that in 2017, the number of invention patent applications in China was 1.382 million, a year-on-year increase of 14.2%. A total of 420,000 invention patents were authorized, including 327,000 domestic invention patents, an increase of 8.2%. The surging momentum of generate’s patent industry, which is rising in quantity and quality, has made the pace from a big intellectual property country to a strong intellectual property country more solid and powerful.

Innovation ability is increasing day by day.

Enterprises have performed very well in the application and authorization of invention patents. "In 2017, the proportion of domestic invention patent applications and ownership in China reached 63.3% and 66.4% respectively, up by 1.6 and 0.9 percentage points from 2016." Bi Nan, Director of China National Intellectual Property Administration Planning and Development Department, said.

In 2017, among the top ten enterprises in China (excluding Hong Kong, Macao and Taiwan) in the number of invention patents granted, there are both "frequent customers" like Huawei, Sinopec, ZTE and BOE, and also new industries like Opal. There are senior strength groups such as Gree Electric, Lenovo and State Grid, and rising stars such as SMIC.

"The ranking of the number of enterprise invention patents granted this time shows the latest progress of China’s innovation-driven development strategy from one aspect, and it is characterized by diversity in one word." Liu Haibo, a researcher at the Institute of Science and Technology Strategy Consulting of China Academy of Sciences, told the Economic Daily reporter that, just as biodiversity is a symbol of the ecosystem, the diversity of the top ten enterprises in the list indicates that China’s innovation ecosystem is becoming more and more active, more resilient and has greater potential for sustainable innovation.

Su Ping, dean of the Intellectual Property College of Chongqing University of Technology, believes that the list directly reflects that the innovation ability of some enterprises in China has been continuously strengthened, and the status of enterprises as the main body of intellectual property creation and application has become increasingly stable. Counting the ranking of invention patents granted by domestic enterprises, Huawei, ZTE, BOE, Opal and Lenovo are all technology-based enterprises focusing on communication and electronic information technology, occupying five seats in the top ten. At the same time, ZTE, BOE and Huawei are among the 100 innovative enterprise clusters in the "2017 Global Innovation Index" published by the World Intellectual Property Organization, which shows that Chinese enterprises are gradually occupying a dominant position in these fields.

In addition, as a representative of R&D-intensive enterprises, SMIC International Integrated Circuit Manufacturing (Shanghai) Co., Ltd. also ranks among the top ten with 862 invention patents. "This shows that enterprises can catch up as long as they attach importance to innovation and increase patent protection and application." Su Ping said.

Compared with the ranking of invention patents granted by domestic enterprises (excluding Hong Kong, Macao and Taiwan) in 2016, it is not difficult to find that the top five enterprises in the past two years are basically the same. In this regard, Su Ping explained that patent creation, application and authorization are closely related to the law of innovation and development. On the one hand, under the influence of the new round of global innovation competition leading paradigm, the above-mentioned enterprises put intellectual property strategy at the core of enterprise development and invested a lot of R&D power to carry out innovation, so the number of authorizations is naturally large; On the other hand, enterprises that have tasted the sweetness in the patent market pay special attention to patent layout, take patent protection measures for their innovative achievements, and add value to enterprises through patent operation, forming a virtuous circle from innovation to patent to benefit.

Pay more attention to quality improvement

Different from the past, this time China National Intellectual Property Administration only published the ranking of domestic enterprises’ invention patents granted, but did not publish the ranking of domestic enterprises’ invention patent applications. In this regard, China National Intellectual Property Administration spokesperson Hu Wenhui explained that the purpose of adjusting the disclosure content of patent statistical data is to further improve the quality of patents and give full play to the innovation-oriented role of patent statistical indicators.

"This really sends us a signal that China will pay more attention to the quality of patent applications." Su Ping said that judging from the proportion of China’s patents granted in the world, China’s patents have won in number and become a veritable intellectual property power, and the next step is to improve the quality. It is reported that by the end of 2017, the number of invention patents in China (excluding Hong Kong, Macao and Taiwan) totaled 1.356 million, and the number of invention patents per 10,000 population reached 9.8.

"The level of patent creation has progressed steadily, thanks to the fact that in recent years, China has paid more attention to improving the quality and efficiency of intellectual property while paying attention to improving the quantity of intellectual property." Bi Nan said that since the 18th National Congress of the Communist Party of China, China National Intellectual Property Administration has put forward the concept of "quality wins, quantity distribution" and started to implement the patent quality improvement project. Focus on the whole chain of patent work, formulate a series of targeted measures around important links such as patent application, agency, examination, protection and application, and improve the quality of patents with multiple measures. Improve the efficiency of patent application, create a good patent protection environment, and promote high-quality creation and high-value patent implementation.

"Eye of the Sky" sounding, "Dragon" going into the sea, "Kunlong" flying successfully for the first time, and high-speed rail running out of "China speed" … Driven by a series of measures, China’s patent creation has formed a good situation in which the quantity and quality have risen together. Among the 35 technical fields classified by the World Intellectual Property Organization, in 2017, the number of domestic invention patents was higher than that of foreign invention patents in China by 30. "But we should also be soberly aware that there are not many foreign enterprises applying for invention patents in China, and there is still a long way to go from quantity layout to quality victory." Su Ping said.

"The content of patent statistics released this time has changed significantly, which really conveys the judgment of the government authorities on the current patent situation and the signal of the next policy adjustment." Liu Haibo believes that at present, creating high-quality core patents should be listed as the top priority of China’s patent work, but the transformation process from winning by quantity to improving quality is long and even more painful.

Some areas need to be strengthened

Although Chinese enterprises have grown rapidly in the fields of mobile communication, electronic information technology and chemical industry, we should also see that there is still a gap between the patent layout in some fields and that in foreign countries. "At present, in the five fields of optics, medical technology, engine, audio-visual technology and transportation, the number of foreign invention patents in China is higher than that in China." Bi Wei said that from the perspective of maintaining the number of invention patents for more than 10 years, the number in China is still less than that in foreign countries in 29 technical fields, and it is still necessary to continue to vigorously cultivate high-value core patents in the above fields.

"If mobile communication, electronic information and other industries invest heavily in research and development, they can present technological achievements in a short period of time. In order to make a breakthrough in high-end manufacturing, we must make long-term breakthroughs in core technologies such as bearings, gears, reducers and transmission devices. " According to Tang Heng, director of Jiangsu Intellectual Property Research Center, the reason why China lacks patent layout in high-end equipment such as optics, transportation and engines is because of the lack of innovation accumulation in these fields.

Patents are strategic weapons for enterprises to control the market and participate in competition, while vigorously developing high-end equipment is the only way to promote the transformation and upgrading of traditional manufacturing industries. Therefore, patent layout in the field of high-end equipment plays a vital role in revitalizing China’s real economy. "Just like strategic weapons need long-term pre-research, high-input research and development, high-intensity testing and high-quality production, cultivating high-quality core patents in the fields of engines and transportation requires creative courage to climb and persistent innovation." Liu Haibo said.

A very important reason for the short patent maintenance time is that the quality of the patent itself is not high, the application value is low, and the demand for technology market is shrinking. "From the perspective of maintaining the number of invention patents for more than 10 years, we do have a certain gap with foreign countries, which also shows that the quality of patents in these technical fields in China is not high enough." Su Ping suggested that in the future, Chinese enterprises should increase investment in R&D, plan the patent layout in advance according to the development direction of technology, and provide good technology from the source of innovation. At the same time, we should further strengthen the ability of patent application and improve the level of patent industrialization.

"In the process of patent examination, the state should moderately shift the focus to the patent maintenance time. Enterprises should also pay more attention to the second or even multiple research and development of patents, aim at the forefront of the industry, constantly integrate business needs with patent creation, and give them stronger market functions, so as to effectively extend the patent maintenance time. " Tang Heng said. (Reporter Li Peng Da)

Trump "shabu-shabu" WHO CNN repeatedly asked 12 questions to question the US government.

  The picture shows the White House epidemic briefing. (Photo: Associated Press)

  [Overseas Network April 16th | All Time Zones of War and Epidemic]US President Trump announced on the 14th local time that the United States suspended funding for the World Health Organization (WHO) and officially launched an investigation into WHO’s response to the epidemic. CNN15 published a commentary on the 15th, asking 12 questions in succession, questioning the measures taken by the US government in epidemic prevention and control, and thought that "the Trump administration’s response to this epidemic crisis needs careful review".

  After Trump announced "no supply", he threw the pot at WHO, claiming that the organization had "mishandled" and "concealed" the COVID-19 epidemic, and that "their mistakes caused so many deaths". CNN believes that this statement is a desperate attempt to divert people’s attention from the "catastrophic behavior" of the US government. The article mentioned that the COVID-19 epidemic was one of the biggest security failures in American history, and suggested that the US congressional oversight committee should immediately review the Trump administration’s handling of this "pandemic" epidemic failure, so that the United States can change its direction.

  The article further lists the specific events of the US government during the epidemic, and uses 12 "question marks" to question them. The specific contents are as follows:

  China reported the COVID-19 outbreak to WHO on December 31, 2019, and the National Security Council of the United States received a notification of the epidemic from the Health Commissioner of the United States in Beijing at the same time. The investigation should determine whether and when the White House received the news, and when it was the first part of the president’s intelligence briefing. For example, the article said that South Korea began screening flights on January 3, while the screening in the United States on January 17 was chaotic, and there were even no such measures in many places with serious epidemics."Why didn’t the US government conduct flight screening?"

  On January 3rd, Gao Fu, director of China CDC, called Robert redfield, director of US CDC to discuss the epidemic situation."What did they discuss? When will the details of the call be conveyed to Trump and the US National Security Council? "

  According to Washington post’s report, US intelligence agencies released confidential briefs on the epidemic emergency in January and February respectively."What is the warning content,Did the intelligence agencies make specific suggestions? "

  Trump now seems to think that the United States has been "misled" by WHO. However, in view of the fact that many other countries are working in high emergency situations,"To what extent does the United States rely on WHO statements and warnings to formulate policies?"

  On January 23, the WHO Emergency Committee did not designate the new pneumonia epidemic as a "global health emergency" but finally announced it on January 30. The US Centers for Disease Control and Prevention has a senior official on the Committee. At the meeting on January 23,"What advice did the official put forward? Does the representative of the United States ask him to declare a global health emergency? Has the position of the US Centers for Disease Control and Prevention been discussed with Trump? "

  In early February, the US Centers for Disease Control and Prevention (CDC) sent out the wrong test boxes in various states in the United States, and some red tape prevented this problem from being solved as soon as possible. Other countries have developed test boxes and conducted large-scale tests several weeks before the United States."What is the reason for the basic mistakes in the CDC?"

  The federal government and many states have been competing for scarce and urgently needed supplies, including personal protective equipment (PPE), ventilators, testing kits and other medical infrastructure."What information does (the US government) use to decide how to assist each state in obtaining these medical necessities?"Trump’s long-term delay in invoking the power of the National Defense Production Law has led the governor and mayor to fight for these missing materials until now."Why did you refuse to use the National Defense Production Law in many key weeks?"

  Up to now, there is no agreed national strategy for "reopening the economy". On the contrary, Trump has an open conflict with many governors."What federal management systems and institutions are (the US government) using to obtain the best information and evidence so as to understand how to conduct and coordinate the actions of the federal government and the state government?"

  CNN believes that all the above are urgent questions to be answered, but the United States failed to take early action, including mobilizing the supply of important commodities and coordinating among national, state and local levels, and it is the responsibility of the House Oversight Committee to solve these urgent problems. The article concludes, "The lives of hundreds of thousands of Americans are in danger, not to mention that the United States should help save the lives of millions of people around the world. Congress must urgently investigate and correct the catastrophic failure of the Trump administration in dealing with this crisis. There is no time to delay. " (Overseas Network/Tommy)

During the month, more than 20 car officials announced the price reduction, and the "price war" in the auto market intensified.

Since March, the price war in the automobile industry has continued to spread.
On March 14th, Dongfeng eπ007, a medium-sized and large B-class car, announced a price reduction of 30,000 yuan as soon as it was listed, and the entry price dropped to 129,600 yuan. In fact, this is just a microcosm of the price war in the automobile market since March 1.
"China Business News" reporter noted that the current price war in the automobile market is still fierce, and the "follow-up" price reduction similar to Dongfeng eπ appears frequently. Take LI as an example. On March 12th, LI announced that the prices of L7 Air and L8 Air, which were listed in February this year, would be lowered by 18,000 yuan.
New energy vehicles, which have become "price butchers", have brought a lot of pressure to fuel vehicles invisibly, and fuel vehicles have followed the pace and started to cut prices. Geely announced the launch of the Spring Car Festival in 2024 a few days ago, and its fuel vehicles such as Xingyue L, Xingrui and Emgrand all reduced their prices, ranging from 3,000 yuan to 14,000 yuan.
With the gradual decline of new energy prices, the fuel vehicle market with relatively slow upgrading and low product intelligence is under serious pressure. In this regard, many people in the industry said in an interview that in 2024, the "scissors difference" between the sales of fuel vehicles and new energy vehicles will be further expanded. Under the background of the gradual increase in the penetration rate of new energy vehicles and the decline in the cost of new energy vehicles, "the same price of oil and electricity" began to appear, and the market share of fuel vehicles may be further reduced.
More than 20 models announced price cuts
In this regard, Cui Dongshu, Secretary-General of the Association, issued an analysis, saying that as of March 15th, the price reduction scale of the automobile market in 2024 was half that of the whole year in 2023, and the price reduction models were mainly pure electric vehicles, plug-in hybrid vehicles and other new energy vehicles, while the price reduction of fuel vehicles was less.
The reporter noted that since March 7, Yantu has launched a replacement subsidy policy of 1 billion yuan, covering all its models of Lantu FREE, New Lantu Dreamer and Lantu Chasing Light, and can enjoy a cash discount of up to 50,000 yuan when buying a car.
Similarly, Chery Automobile also announced that its four major brands Chery, Xingtu, Jietu and iCAR have launched subsidy activities. Tiggo 9, Tiggo 7 Series, Arrizo 8 Series, Exploration 06, Fengyun A8 and other models can trade in up to 30,000 yuan.
It is worth noting that some fuel models of Chery Automobile have also joined the ranks of price reduction. The limited-time comprehensive discounts of Tiggo 5x, Tiggo 3x, Arrizo 5 Series and Oumengda range from 6,000 yuan to 19,000 yuan. At the same time, the Touran brand of SAIC Volkswagen also started a limited-time discount on March 1, and the price of the 2024 Touran four-wheel drive luxury and Touran X four-wheel drive luxury designated models was reduced to 279,900 yuan and 265,000 yuan. In addition, the 2.0T model of Touran family enjoys 2,000 yuan in cash or 3 years of worry-free maintenance.
According to Zhou Wenyu, deputy director of Huiyu Bohua Industrial and Commercial Enterprise Department, some mainstream car companies set aggressive annual sales targets at the beginning of the year. Under the expectation of low sales growth in the industry, "exchanging price for quantity" and seizing the market as soon as possible will still be one of the strategic choices of car companies. "The fast-growing car companies intend to pursue victory and expand market share, while the slow-selling manufacturers are unwilling to lag behind and intend to catch up. At the same time, the downward cost of upstream general raw materials and power battery materials, as well as the economies of scale brought about by the expansion of sales of independent brands, have also provided some car companies with greater room for profit. "
The fuel vehicle market may be further under pressure.
According to the data of the Ministry of Public Security, in 2023, there were 24.56 million newly registered cars nationwide, an increase of 1.33 million compared with 2022, with a year-on-year increase of 5.73%. Among them, the number of new energy vehicles reached 7.43 million, accounting for 30.25% of the number of newly registered vehicles, up 38.76% year-on-year. In 2023, the number of new energy vehicles in China has reached 20.41 million, accounting for 6.07% of the total number of vehicles. Among them, the number of pure electric vehicles is 15.52 million, accounting for 76.04% of the new energy vehicles.
Although the proportion of the total is still low, the growth rate of new energy vehicles is still relatively fast. According to the data of China Automobile Industry Association, the domestic production and sales of new energy vehicles continued to maintain a high growth rate in January and February 2024, among which the sales volume of new energy vehicles reached 1.207 million, up 29.4% year-on-year, and the market share reached 30%.
With this year’s "Government Work Report" proposing to encourage and promote the trade-in of consumer goods and boost the mass consumption of intelligent networked new energy vehicles and electronic products, in the eyes of many insiders, the sales of new energy vehicles will further increase, and the sales of fuel vehicles will be further under pressure.
"With the improvement of new energy penetration rate, as well as in the context of policy support and consumer awareness, the market share of new energy vehicles will continue to grow. Fuel vehicle manufacturers may face increasing challenges and need to maintain competitiveness through technological innovation and cost control. " Jiang Han, a senior researcher at Pangu think tank, told reporters that the development trend of fuel vehicles and new energy vehicles may be divided this year. "Overall, new energy vehicles will continue to maintain a strong growth momentum, while the market share of fuel vehicles will gradually shrink.".
Zhou Wenyu also told reporters that the performance of the automobile market in January 2024 was affected by factors such as season and sales overdraft in December last year. In the long run, as the penetration rate of new energy vehicles continues to rise, the cost goes down, and the level of intelligence continues to improve, the market share of fuel vehicles will be squeezed at an accelerated pace, and the main engine manufacturers with fuel vehicles as their mainstream products will also face the risk of accelerating transformation or being eliminated by the market.
"It is expected that in 2024, various regions will still introduce consumption promotion policies such as trade-in and new energy vehicle consumption subsidies." Zhou Wenyu said that Shanghai, Chongqing and other regions took the lead in opening a new round of automobile consumption subsidies. With the sustained economic recovery and increased consumer confidence, the excess savings accumulated by residents will gradually be converted into related consumption and investment, which is conducive to the growth of automobile sales. "
According to previous analysis, with the rapid increase in the penetration rate of new energy vehicles, the market size of traditional fuel vehicles has gradually shrunk, and the contradiction between huge traditional production capacity and shrinking fuel vehicle market has brought more intense price wars. The scale determines the cost and the survival state of the enterprise, and most manufacturers give priority to keeping their share, which will inevitably lead to further intensification of price competition.
The price war will continue
According to the analysis of the Federation, 2024 is a crucial year for new energy vehicle companies to gain a foothold. The downward exploration of new energy costs and the "same price of oil and electricity" have brought tremendous pressure to fuel vehicle manufacturers. "The upgrading of fuel vehicle products is relatively slow, and the degree of product intelligence is not too high, and it relies more on preferential prices to continuously attract customers."
"With the decline in the price of lithium carbonate, the cost of batteries has decreased, and the cost of building new energy vehicles has decreased; And with the rapid development of the new energy market, the scale effect will be formed, and the products will have more profit space. " The Federation further stated.
Cui Dongshu said that the new characteristics of new energy consumption have brought about obvious brand fission of automobile enterprises. With the gradual maturity of the market segments of various technical lines of new energy vehicles, new models have greatly enriched consumption choices, and the homogenization brought about by the increase of product supply will also intensify competition. This year’s price war is likely to be carried out in a mixed mode of price reduction promotion and positioning exploration, and new energy manufacturers will package more rights and interests to stabilize product prices.
At present, major car companies are accelerating competition on intelligent tracks. According to the White Paper on the Internet of Vehicles in China released by China ICT Institute, by 2025, the smart car market in China will be close to one trillion yuan, reaching 960 billion yuan, accounting for 56.5% of the global market. From 2020 to 2025, the smart car market in China will grow at a compound annual growth rate of 36.9%.
In the face of the trillion-dollar blue ocean, there have been a lot of R&D and investment by car companies in the field of intelligent driving. Take extreme krypton cars as an example. Recently, extreme krypton cars said that in 2024, the personnel size of the extreme krypton intelligent driving department will be expanded to more than 2,000 people, and the extreme krypton intelligent cockpit will also accelerate the iteration. "In 2024, Krypton will continue to invest and lay out the big model, and will develop the self-research ability of seamless combination of traditional voice and big model, with better understanding ability and application based on big model."
Compared with new energy vehicles, due to hardware constraints, fuel vehicles are making slow progress on intelligent tracks. How can fuel vehicles as a traditional industry boost consumption? It may be a good solution to start with optimizing management.
Cui Dongshu recently issued a document suggesting that cities with less than 4 million vehicles should gradually liberalize the purchase restriction of fuel vehicles. "At present, the car ownership in some cities with restricted purchases has lagged far behind other cities without restricted purchases. In the future, there is still huge room for growth in domestic automobile consumption. Among them, the growth potential of small and medium-sized cities and county and township markets is huge, and megacities also have room for improvement in automobile consumption. "
Cui Dongshu further stated that fuel vehicles have to pay trillions of yuan in fuel tax every year, which is not only restricted but also restricted. "In the case that the sales volume of new energy vehicles has reached 35%, we should consider the same car and the same power, stabilize the normal consumption of fuel vehicle users, and achieve comprehensive and sustainable growth of automobile consumption."
In fact, despite several rounds of price wars, the current competition pattern in the automobile market is still unclear. Zhou Wenyu told reporters that under the background of high base, new energy car companies are gradually facing the dual pressures of overcapacity and slowing demand growth. Enterprises in the industry will still experience brutal competition, and tail car companies are at risk of going out.
"At the same time, the continuous expansion of the’ Huawei’ series of cooperative brands highlights the changing pattern of the automobile industry chain." Zhou Wenyu said that the current cross-border automobile industry of science and technology enterprises indicates that the automobile industry chain is moving from a closed network to an open one. Traditional automobile host manufacturers and suppliers are facing challenges from new players, and the competition pattern is far from clear.
Source: China Business Network
Reporting/feedback

Stumbling at the historical turning point: the prospect of emerging market economy in 2020-the truth

 "I would cross the Yellow River, but ice chokes the ferry, the Taihang Mountain will be covered with snow." Looking back, ten years is a cycle. According to IMF (International Monetary Fund) forecast, the economic growth rate of emerging markets will fall to the lowest level since 2009 in 2019, and the historical logic of mean regression has ignited the market’s longing for its bottoming out. Looking forward, ten years is a turning point. Although the short-term impact may dissipate, the long-term qualitative change has been formed, and it is doomed that history will not simply repeat itself.

Looking forward to 2020, there are three historical turning points on the road of recovery in emerging markets.

First, the main engine entered a period of fatigue.While China’s economy insists on "slowing down and improving quality", the Indian economy still lacks the steady power to take over, and the weakness of "BRICS" will restrain the rebound of emerging markets.

Second, the main mode hit the ceiling.As the debt cycle enters the second half, the debt-driven growth that emerging markets rely on is unsustainable, while the narrow policy space and rising populist risks are increasing the threat of debt deflation.

Third, the old dividend has become a new weakness.As the external pillar of emerging markets, economic globalization is easy to retreat but difficult to advance, and emerging markets will face multiple risk shocks.

Stumbling on the three historical turning points, the growth rate of emerging markets is expected to continue to fall near the low level of the ten-year cycle in 2020. The sharp contrast between the cold growth rate and the hot expectation will constitute the biggest risk of investment in emerging markets in 2020. However, conservatism is still the most important value factor, which will give China economy a comparative advantage and capital favor.

One of the inflection points: As the main engine of emerging markets, the BRICS countries have entered a period of weakness.

Different from the European Union, North America and other regions, the so-called "emerging market" has always been a loose concept, with neither close external geographical connection nor strong internal organizational coordination. Thus, since the beginning of this century, the growth path of emerging markets has been dominated by the BRICS countries relatively independently. Although there are a large number of other emerging market economies, it is difficult to have an important impact. From 2000 to 2018, the contribution rate of BRICS countries to the added value of GDP in emerging markets was about 59%. During this period, the correlation coefficient between the GDP growth rate of BRICS countries and the overall growth rate of emerging markets was as high as 0.93. In 2002, 2008, 2011, 2016 and other years, although the growth rate of other emerging economies has experienced ups and downs, the overall performance of emerging markets is still similar to that of the BRICS countries. Therefore, the trend of BRICS countries directly determines the ups and downs of emerging markets, and the collective weakness of BRICS countries will also constitute the growth turning point of emerging markets.

Looking forward to 2020, this turning point is coming. As the old engine of emerging markets, the current trend of "slowing down and improving quality" of China’s economy continues, and the desired economic growth rate in the future is expected to be steadily lowered, and the demand for other emerging economies will also be weakened.

As a new engine of emerging markets, the Indian economy may not be able to take over smoothly in the short term. On the economic level, since the second quarter of 2018, India’s economic growth rate has declined quarter by quarter, and the unemployment rate has risen to an all-time high, and it is difficult to find signs of improvement in many leading indicators in the near future. This phenomenon shows that in the case of weak structural reform and slow growth of labor productivity, it is difficult to sustain the consumption growth achieved by consuming household savings too quickly, and the demand-side stimulus of Modi government failed to enhance the long-term motivation of India’s economy. On the financial level, among the major emerging market economies, the deterioration of Indian government debt and fiscal deficit is second only to Brazil. At the same time, since the second quarter of this year, the expansion of the trade deficit, the withdrawal of international capital and the fluctuation of the local currency value are jointly weakening India’s foreign debt solvency, so the debt risk is rising.

Except India, due to low investment, high unemployment, high policy uncertainty and geopolitical conflicts, the economic growth rates of South Africa, Brazil and Russia may be slightly restored in 2020, but the rebound space is narrow. Generally speaking, 2020 will be a year in which the BRICS countries continue to be tired. In this context, it is unrealistic to hope that other emerging economies will drive the strong recovery of emerging markets, and the future growth rate of emerging markets may be weaker than market expectations.

The second turning point: as the main mode of emerging markets, debt-driven hit the ceiling.

Looking back over the past decade, the subprime mortgage crisis and the European debt crisis forced the debt risks of developed economies to be partially cleared. In contrast, since 2008, emerging markets have become increasingly dependent on the debt-driven growth model, and the debt intensity of economic growth has increased significantly. According to the statistics of BIS (Bank for International Settlements) and IMF, from the first quarter of 2008 to the first quarter of 2019, the proportion of non-financial sector liabilities in GDP in emerging markets increased by more than 72 percentage points, which was 57 percentage points higher than that in developed economies. From 2008 to 2019, the median external debt/export of emerging markets also rose from 100% to 160%. While the macro leverage level is high, by 2020, the duration of this round of debt expansion in emerging markets will also approach the historical average of the debt cycle length (15 years). As a result, in the future, emerging markets will move towards the downward stage of the debt cycle with a high probability, actively or passively, and the debt-driven growth they rely on is expected to hit the ceiling of the economic level. Under this trend, due to the limited room for "leverage" in the non-financial sector, although emerging markets will increase monetary easing with the global interest rate cut tide, the materials used to boost the real economy will be marginally attenuated, making it difficult to independently support the steady rebound of the economy.

Facing the limitation of economic ceiling, it will be an inevitable choice to deal with the downward pressure of the economy by taking active fiscal measures to ease the currency. However, the ceiling of policy and politics will be highlighted, which will constrain the active finance. From the perspective of policy space, according to IMF data, from 2009 to 2019, the average value of general government debt /GDP in major emerging market economies has climbed from about 39% to over 46%; On this basis, compared with 2009-2018, the average fiscal deficit ratio of major emerging market economies is expected to rise by 1.8 percentage points in 2019-2024, which is expected to accelerate the accumulation of public debt. As a result, the extra space available in fiscal policy to stimulate the economy is relatively limited, and the endogenous contradiction between government budget balance and economic growth may be further aggravated.

Judging from the political situation, populist forces have become a real risk in emerging markets. Since Brexit in 2016 and Trump was elected president of the United States, the rise of populism in developed economies has attracted attention, but in fact, the rising tide of populism in emerging markets has quietly surpassed the former in depth and intensity.

At the deeper level, taking Europe with more internal comparability as an example, in the second quarter of 2019, among the top ten European populist parties, emerging economies occupied eight seats. Among them, Hungary, Poland and Czech Republic, which are regarded as the recovery engines of Central and Eastern Europe by the market, rank first, second and third respectively. The developed economies only have two seats (Italy and France).

On the strength level, in emerging markets in Latin America, such as Brazil, Chile and Venezuela, populism has shown an extreme trend. Based on this, following the developed economies from 2016 to 2019, the long-accumulated populist risks will break out in emerging markets such as Central and Eastern Europe and Latin America in 2020. This risk will disturb the internal policy rationality and resonate with the external geopolitical game, which is expected to increase the sovereign credit risk of emerging markets and the financing cost of financial institutions, thus amplifying the possibility of triggering "debt deflation".

The third turning point: located in the era of global chaos, the old dividend is becoming a new weakness.

Based on a longer-term historical perspective, in the 30 years since the end of the Cold War, the relatively stable international order and the continuous promotion of global economic integration have built a long-term fast lane for the development of emerging markets. However, due to the drastic changes in the international environment in 2018-2019, the external dependence of emerging markets has degenerated from dividends to weaknesses, and will be manifested from three levels in 2020.

First, the peak of trade friction impact is approaching.Although the global trade game has shown signs of easing since September, the cost accumulated by the previous friction still needs to be cleared. In this round of shocks, the pressure on emerging markets is expected to be higher than that on developed economies. According to the latest forecast of IMF, the negative impact of trade friction on global economic growth in 2020 will rise to the peak in 2018-2023. Moreover, taking G20 members as an example, the negative impact of trade friction on the economic growth of emerging markets in different scenarios is stronger than that of developed countries; The worse the game situation, the wider the impact channels, and the more obvious the difference in intensity. Even if the trade friction measures announced after August 2019 are removed, the above-mentioned difference in impact intensity is expected to reach about 0.24 percentage points.

Second, commodities continued to slump.Against the background of weak global aggregate demand and slowing industrial production, it is difficult for the price centers of energy commodities such as oil, coal and natural gas to rise in 2020. Although short-term supply and price fluctuations may occur due to geographical conflicts, meteorological disasters and other factors, it lacks continuous assistance to the income of energy exporting countries. At the same time, base metal commodities will return to weakness. According to the IMF’s forecast, the global base metal price index will end the rebound in 2019 and turn to a marginal decline of 6.2 percentage points in 2020. Considering that 67% of developing countries’ economies are dependent on commodity exports, the international market downturn in 2020 is expected to drag down the recovery of emerging economies.

Third, the shadow of currency risk is hard to disperse.Although the loose stance of the Federal Reserve in 2020 is expected to lead the US dollar index to decline gradually, it does not mean that emerging market currencies can sit back and relax. The experience in 2019 shows that geopolitical conflicts have a higher influence on the currency trend of emerging markets in the short term than the Fed’s interest rate cut process. With the long-term global trade game and geopolitical conflict, emerging market currencies may encounter staged and structural risk shocks in 2020, and thus move towards performance differentiation. According to our calculation, if the above shocks occur, the ten most dangerous currencies in emerging markets are: Venezuelan Bolivarian, Argentine peso, Indian Rupee, Brazilian real, Turkish lira, South African rand, Indonesian rupiah, Malaysian ringgit, Belarusian rouble and Vietnamese dong.

There is no shortcut to crossing the turning point, and the emerging rebound may be late.

Facing the three historic turning points, emerging markets need to accelerate structural reforms internally; In order to adapt to the new global trade system of intra-industry trade regionalization and inter-industry trade globalization, it is necessary to expand its opening to the United States and actively build regional trade agreements with other manufacturing powers. There are no shortcuts to these countermeasures, let alone a year’s work.

Based on the above analysis, we can draw the following three inferences.

First, the strong rebound in emerging markets is expected to be late.In 2020, emerging markets will still be in the long-term process of crossing the historical turning point, and the growth rate is expected to continue to fall near the low level of the ten-year cycle. As a whole, it is difficult to see a sustained and strong growth rebound.

Second, expected reversal will be the biggest risk.In 2020, under the pressure of low interest rates and asset shortage in developed economies, international capital will be forced to seek high returns from emerging markets, thus enhancing its imagination of bottoming out in emerging markets. However, once this expectation seriously deviates from the weak growth of emerging markets, the hard facts will eventually shatter the illusion. Fast forward and fast out of capital will touch the debt risk and exchange rate risk, and the roles of "honey" and "arsenic" will switch more frequently.

Third, robustness will be the most important value factor.Under the threat of expected reversal, "steady low growth" is more valuable than "fragile high growth" As analyzed above, from the perspectives of debt risk, exchange rate risk and populist risk, compared with other emerging markets such as India, Central and Eastern Europe and Latin China,’s economy "slowing down and improving quality" still has a high stability, which is expected to gain value advantages and capital favor.

(Cheng Shi is chief economist, managing director and head of research department of ICBC International, and Qian Zhijun is senior economist of ICBC International)

Shandong University | The key to widening the economic gap between the north and the south of China lies in the gap in innovation ability.

[Editor’s note]

On October 16th, the 17th (2019) annual academic conference of Shanghai social sciences celebrated the 70th anniversary of the founding of New China, and the sub-forum of Shanghai Economic Achievements: Theoretical Summary and Prospect was held in Shanghai Normal University. The experts and scholars attending the conference gave speeches and discussed many topics on China’s economic development in the past 70 years. At the forum, Huang Shaoan, Dean of the Economic Research Institute of Shandong University, delivered a speech with the theme of "Differences in Informal Institutions and the North-South Divide of China’s Economy". Huang Shaoan made a deep comparison between Shandong’s economic development and Guangdong’s and Jiangsu’s, and expounded why Shandong lacked innovative enterprises and talents, and what was the root cause of Shandong’s economic decline?

The following is the full text of Professor Huang Shaoan’s speech compiled by The Paper for readers.

Vision china data map of Shenzhen night scene

Recently, we are doing a project, comparing Shandong with Guangdong and Jiangsu in the south. In the process of doing it, we found an important phenomenon. The imbalance of regional economy in China has rapidly changed from the imbalance in the eastern, central and western regions in the past to the imbalance between the north and the south and the gap between the north and the south in recent years, and the gap is obvious.

1. After the 2008 financial crisis, the substantial economic gap between the north and the south of China has begun.

In terms of growth rate, before 2014, there was little difference between the growth rates of the north and the south, but since 2014, the GDP growth rate of the northern region, including the northern leader Shandong Province, has been divided from that of the southern region, and the growth rate of the southern region is significantly higher than that of the northern region. Statistical data began to appear in 2012, and it began to be obvious in 2014. It is obvious that this speed is very fast this year.

In the first half of 2019, the GDP of 31 provinces and cities, excluding Hong Kong, Macao and Taiwan, increased, with the national average growth rate of 6.3, and the average growth rate of 16 provinces and cities in the south was 7.38, only 5.9 in Shanghai, 6.2 in Chongqing and 5.3 in Hainan, which was lower than the national average growth rate, but Shanghai had a large base and the kinetic energy of growth rate changed faster. The average growth rate of 15 provinces, municipalities and autonomous regions in the north is 5.63. Only Beijing 6.3, Henan 7.7, Hebei 6.62, Shanxi 7.2 and Ningxia 6.5 are slightly higher.

The total GDP of the north is about 172,354 trillion yuan, accounting for only 38% of the national total, while the south accounts for 62%. This is the data for the first half of 2019. One in the north is Beijing-Tianjin-Hebei, and the other is Shandong plate, which was more than 7 trillion last year. One province in Shandong is equal to three provinces and cities of Beijing-Tianjin-Hebei.

From the perspective of personal income tax, the gap between the north and the south is even greater. The income tax paid by residents of Zhejiang, Jiangsu and Guangdong is obviously higher than that of Shandong, which represents the largest economy in the north, and the difference is too far.

1. The gap actually started after the 2008 crisis.

What I saw before is a statistical difference, as if 2014 was a watershed. In fact, the substantial gap did not start in 2014, but should have started when the world financial crisis spread to China in 2008. The substantive difference lies in the different measures to deal with the economic crisis. After 2008, Guangdong, especially Shenzhen, Guangzhou and Dongguan, underwent painful economic restructuring. When the economic crisis came, some resource-consuming, labor-intensive and low-end exports were forcibly eliminated by relying on the market mechanism.

Of course, the statistics don’t show it. At that time, the growth rate of the north and the south didn’t show it. From which aspect can it be seen?

Economic structural transformation. After the financial crisis in 2008, Guangdong, Zhejiang, Fujian and Jiangsu, especially Guangdong, implemented economic structural transformation, which is what we call the transformation of old and new kinetic energy. The transformation of Shenzhen, Dongguan and Dongguan was particularly successful. Let’s go to Dongguan again, and the economic transformation turned particularly well. At the same time, Shandong is actively undertaking the transfer of relatively backward industries such as Guangdong and Zhejiang. Although the economic growth rate will be relatively high in the short term, it actually puts itself in the position of accepting the transfer of backward production capacity. This is the gap.

Therefore, the substantive gap began in 2008, but it can’t be seen from the economic aggregate and GDP growth rate. The growth rate of GDP, the total GDP and the transformation of industrial structure are all important, but they are not the core.

2. The core and key gap in the widening economic gap between the North and the South is the gap in innovation ability, not the industrial structure.

In fact, the industrial structure itself does not matter whether it is good or bad, only whether it conforms to the quality structure of its own production factors, and there is no problem of upgrading the industrial structure, only the problem of adjustment and optimization. However, the industry has the problem of upgrading, that is, improving quality. There are no backward industries, only backward industrial technology and management. Not every place should raise the tertiary industry to more than 70%, which is contrary to the most basic principles of economics. The structure of production factors determines the industrial structure, and the quality of production factors determines the quality of industry. The key is the level of each industry, the technical content and the management innovation ability. The same manufacturing industry, Guangdong and Hebei are different, the same agriculture, Shandong and Shanxi are different.

First of all, there is a gap in total R&D investment. Shandong has the highest investment in the north, ranking third in the country, and the total amount is also the third in the country, but it is not in the same file as Guangdong and Jiangsu, both of which are more than 200 billion, and Shandong is more than 100 billion. There is also the intensity of investment, which accounts for the proportion of GDP. It is also obvious that the south is higher than the north. The total amount of this investment is not the most critical. The most critical thing is the input-output ratio. Look at Shandong, which is the most invested in the north, with more than 100 billion yuan, equivalent to 70% of Guangdong’s investment. The difference of 30% should also be far away, but the biggest gap is not in these places. The key is the output after investing these money. According to the patent index, this gap is two or three times the order of magnitude, and it is invested in Guangdong, Zhejiang and Jiangsu.

Why is the difference between input and output so far? We’re going to analyze the reasons.

Another is the innovation level of representative cities, and Shandong represents the north. Beijing is a very special city. Generally speaking, the cities in the north are obviously behind Shenzhen, Guangzhou, Nanjing, Wuhan, Tianjin and Hangzhou, and even worse than Shanghai. In this North-South economic structure, the gap between the economies of big cities and megacities as a percentage of GDP can basically determine this gap in innovation capacity. To tell the truth, the real innovation is mainly in big cities, and there is little innovation in GDP below the prefecture level.

Compared with the south, there is a shortage of innovative scientific and technological talents in the north. In fact, there are not necessarily fewer people with titles in the north than in the south. There are many academicians, Changjiang scholars and so on in Beijing. However, the scientific and technological talents in the south pay more attention to innovation itself. You may think that Shandong is a big province of marine economy, and you may think that Shandong’s marine scientific and technological talents are the first in the country. In fact, the number of academicians in the field of marine science and technology is the first in the country, and others are far behind Guangdong. Moreover, the total GDP of Shandong’s marine economy has lagged behind that of Guangdong. In 2018, Guangdong was 1.9 trillion and Shandong was 1.6 trillion. It is not that academicians are not important, but if there are only academicians, it is difficult to give full play to the role of academicians.

Therefore, the economic gap, statistical data is not the most fundamental, the most fundamental is the ability to innovate.

Second, the economic gap between North and South lies in the different preferences of system and economic development model.

Economy determines whether a region’s long-term economic development has enough development momentum, whether it can catch up with the new era and whether it can move forward further. Culture also plays a great role in it. The reason for the economic gap between North and South lies in the different preferences of system and economic development model. The relatively prominent official-oriented culture and hierarchical concept in the north are naturally close to the planned economic system and state-owned enterprises, while they are not close to or even contradictory to the equal competition, legal consciousness, scientific consciousness and innovation consciousness of the market mechanism.

First, the North prefers state-owned enterprises and planned economic systems, including officials of local governments at all levels. They are handy in using and managing state-owned enterprises and planning means, and the government is generally strong and controls and allocates resources.

Second, both state-owned enterprises and private enterprises are heterogeneous between the north and the south. The whole country, Shandong itself and the central leaders all feel that Shandong’s private economy is underdeveloped and state-owned enterprises dominate the world, but Shandong’s contribution to GDP, the number of private enterprises and the number of top 500 enterprises in China are among the best in the country.

Why does everyone think that Shandong’s private economy is underdeveloped? This is the economic mystery of Shandong. Because: Shandong’s state-owned enterprises are very state-owned, and Shandong’s private enterprises are also like state-owned enterprises. We checked the state-owned enterprise groups in Shandong Province, including some large subsidiaries, among which the senior executives, party secretary, general manager and deputy prime minister are basically officials of the former party and government organs sent down by the organization department, while managers of state-owned enterprises in Guangdong, Zhejiang and Fujian, such as Wang Shi and Dong Mingzhu, thought they were bosses of private enterprises, but in fact they were bosses of state-owned holding enterprises, and many of them were entrepreneurs screened out by the market. It is not that officials of the party and government organs will not be enterprises, not necessarily, but the overall big data analysis, this ratio is so far away, it is hard to say that the party and government organs will not be brought to the enterprise.

The gap between private enterprises. Private enterprises in Shandong are different from those in the south. Private enterprises in Shandong should first establish good relations with state-owned enterprises or cadres of party and government organs. In 2017, we conducted a questionnaire survey and listed 14 factors to do well in business management. We interviewed these private enterprise bosses randomly. As a result, two-thirds of the business owners ranked good relations with state-owned enterprises and party and government organs in the top three, and one-third ranked this factor in the first place. In addition, the industrial institutions of private enterprises are developed around state-owned enterprises, and they are highly dependent on state-owned enterprises. So why does Shandong’s statistics show that the private economy is very developed, but the central leadership, Shandong itself and the people of the whole country all think that Shandong’s private economy is underdeveloped.

Third, different institutional preferences lead to different economic development models.

Guangdong is a "private economy+foreign capital economy+market-oriented state-owned economy+relatively weak county economy". Of course, this county is a non-Bay area, because most of the developed counties are in the Bay Area, and many of them are not counties, but prefecture-level cities, such as Dongguan. Another is "super-large urban economy+security government". Therefore, all kinds of enterprises in this economy have strong vitality, great market mechanism, high economic development, strong innovation ability, strong government power, especially strong financial resources, but their scope of action is small, which is equivalent to security.

Zhejiang is a "private economy+market state-owned economy+strong county economy" and a strong "urban economy+security government". Jiangsu is a little different from Zhejiang and Guangdong. Jiangsu is a "collective economy+foreign capital economy (Suzhou is typical)+civilian economy+market-planned state-owned economy+strong county economy" and a relatively strong "urban economy+nanny government". Nanny is a service management type.

Shandong is "planned state-owned economy+quasi-state-owned private economy+strong county economy+weak big city economy", plus Confucian cultural or paternalistic government, what is the difference between paternalistic government and nanny government? Everyone is in charge of the enterprise. The nanny-type government provides services and does not make decisions. The parent-type government is management-oriented and must listen to me.

Fourth, the difference between the north and the south in urban formation mechanism and function. There is a big difference between the cities in the south and those in the north. The cities in the south, especially the central cities, have developed, and the surrounding areas of the cities have been driven. The surrounding areas of Shanghai, Hangzhou, Shenzhen and Guangzhou have all been driven, including some cities in the middle, such as Changsha, which can drive the surrounding areas, that is, the development of the central cities and the development of the surrounding areas, that is, the growth pole of the regional economy. This growth has a very strong polarizing effect. The big cities in the north, no matter how big they are, are very developed inside the city, but outside the city gate, they are relatively poor places, such as Beijing, Tianjin and Xi ‘an. According to the statistics of per capita GDP and per capita income between the northern and southern cities and the surrounding areas, the gap is very large. That is to say, the cities in the north don’t transmit positive energy to the surroundings, but absorb resources from the surroundings. To put it bluntly, they are "vampires". In other words, all kinds of resources are piled up together, which is a "resource pile", not an economic organism, while the cities in the south are an economic organism. The central city and the surrounding areas provide positive energy to each other and promote each other. This is one of the very important reasons for the gap between the north and the south.

Fifth, the cultural differences between North and South. Where are the root causes of the differences between institutional preference and economic development model? Just now, I talked about formal institutional differences, such as state-owned enterprises and planned systems. Are there any differences in informal systems? Shandong represents the north, with relatively prominent official standard and hierarchical concept, and relatively prominent relationship culture, fellow villagers culture, dock culture and circle culture. It is a common phenomenon that China worships officials and ranks the officials, but Shandong is the worst. Relevant data show that since the reform and opening up more than 40 years ago, the number of college students taking the civil service examination and being admitted to universities across the country has been the largest in Shandong. Second, the number and proportion of civil servants in universities in Shandong Province are also the largest in China every year.

At the same time, the relatively prominent concept of official standard and hierarchy naturally leads to the unreasonable flow of elites (excessive flow to party and government organs), which also affects the innovation of local talents and the innovation of foreign talents. It is difficult for people outside Shandong to start a business in Shandong. It is no problem that Shandong is very hospitable, but it is relatively difficult for foreigners to start a business in Shandong. In Jinan and Qingdao, Shandong, the daily working language is basically Shandong dialect, which shows that there are not many foreigners. As a southeast coastal area, the data of the total population flow and structure over the past 40 years can also confirm our judgment.

3. Why didn’t these cultures and systems lead to economic backwardness in the north in the past?

Some people may say that the northern region, including Shandong, used to be the same culture, system and people. Why did it not lead to economic backwardness in the past few decades? Didn’t Shandong’s economy grow very fast in the past, and once ranked first in the country? How to explain it? It turns out that our economy is mainly resource-consuming, environment-polluting, labor-intensive, and a large number of low-end products are exported. The key is resource consumption. Moreover, there are still many resources under the original state planned economy system. Therefore, mandatory planning system, relationship culture, state-owned enterprises, etc., are not necessarily at a disadvantage for economic extension and extensive growth, and some aspects are even dominant, such as obtaining more central resources through relationships and mobilizing various resources on a large scale.

But now it’s different. The economy needs to transform, abandon these traditional growth modes, and promote economic development by innovation. Innovation mainly depends on talents, and talents mainly depend on systems. Informal systems have a great influence on local economic development. What are the main talents introduced from Shandong? Many talented people with titles have been introduced, including academicians. Looking at the statistics, it’s pretty good. A few days ago, Qingdao introduced a large number of academicians in their 80s, and Shandong University also introduced a foreign Nobel Prize in Physics in their 80s. This index system identifies and evaluates talents according to their titles. Major officially introduced talents can only be identified in this way. However, there is a lack of market mechanism to attract, identify, use and treat talents (including respect). This is a hard gap.

It is a fact that the gap between the north and the south is widening. If the concept and system are not changed, the gap between the North and the South may widen further. We also don’t want this gap to widen further.

[Author Huang Shaoan is a professor at Shandong University and dean of the Economic Research Institute of Shandong University. This article has been revised by the author]