China's NDRC and transport ministry launch logistics network plan targeting 13.1% cost-to-GDP by 2030

The Analysis

The document has a date and a target year, so let me start with both. On 28 August 2026 the National Development and Reform Commission and the Ministry of Transport jointly issued the Logistics Network Construction Implementation Plan. It was presented by Li Chao, deputy director of the Commission's policy research office and its press spokesperson, at the Commission's August press conference. The plan deploys 17 key tasks across seven areas, and its headline figure is a target: bring the ratio of total social logistics costs to GDP down to 13.1 percent by 2030, which the plan describes as 0.8 percentage points below the level at the end of the 14th Five-Year Plan. Around 1,000 major freight multimodal nodes are to be upgraded during the 15th Plan period, and the first 300 are already being pushed forward.

Strictly speaking, an implementation plan is not a statute and it is not an administrative regulation. It is a policy instrument issued jointly by two departments, it sits below both of those in the hierarchy, and what it does is deploy tasks and set targets rather than impose duties on anybody outside government. I make that point because it decides how you should read every number in it. An obligation-bearing instrument tells you what you must do and what happens if you do not. This one tells you where the obligation-bearing instruments are going to point. The ones that will actually touch you are the implementing measures, the local plans, the project lists and the application notices that follow, and none of those are in this document.

My observation, having read a good many plans of this shape, is that the error is always the same and always made in good faith. Companies read the target as though it were a rule about them, build an internal KPI out of a macroeconomic ratio, and then spend two years reporting against a number that was never theirs to report against. The disciplined reading is different: this is the shape of the projects, the funding windows and the facility network that will exist by 2030, and the question worth asking in the fourth quarter of 2026 is where you should be standing while it gets built.

Let me take the 13.1 percent apart before anyone builds anything on it, because the denominator does more damage than the numerator. The ratio is total social logistics costs over gross domestic product. The numerator is the whole economy's logistics spending, across every sector and every stage, including the transport, warehousing, handling and management costs of everybody from a steel plant to a corner shop. The denominator is national output. Your company does not appear in either number, and there is no allocation mechanism that delivers a slice of it to you. If your own logistics cost runs at, say, 4 percent of your revenue, that 4 percent and this 13.1 percent are not two measures of the same thing with a gap to be closed. They are two different ratios with different numerators and different denominators, and putting them side by side produces a sentence that sounds meaningful and is not.

There is a second number in the plan that deserves the same treatment, and here I need to flag my own arithmetic. The plan states that the 2030 target is 0.8 percentage points below the level at the end of the 14th Five-Year Plan. Adding 0.8 to 13.1 gives 13.9 percent as the implied baseline. That addition is mine, not a figure the plan quotes, and I flag it because an implied baseline gets repeated as a published one within about a week of any announcement like this. If you are going to use 13.9 percent, use it as an inference from two published numbers and label it that way.

Now the seven areas, because which ones matter to you depends entirely on what you move and where. The plan covers promoting logistics hub construction and interconnection; raising the carrying capacity of logistics corridors; filling gaps in urban and rural logistics facilities; densifying the international logistics facility network; promoting digital and green development of logistics facilities and equipment; improving the sharing and use of logistics information; and stepping up policy support and guidance. Seventeen tasks sit under those seven headings. For an importer or exporter, the pressure points are not evenly distributed. Hub construction and corridor capacity decide what the trunk network looks like. The international facility network heading is the one that touches port-side and cross-border flows. Digital and green development is where equipment and emissions records start to matter. Policy support and guidance is where money, land and approvals live, and it is the heading that turns the other six from description into something you can apply to.

On the main arteries, the plan is specific enough to quote. It will implement a railway freight network project, accelerate the missing sections of the riverside expressway, push capacity expansion and smart upgrading on busy sections of the national expressway network, and support air cargo companies to set up bases at airport-type national logistics hubs. Those are four concrete programmes, and each of them has a geography. A corridor programme is not a national average; it is a list of segments, and whether your freight touches one of them is a question about your lanes, not about the plan. I have not seen the segment lists and I am not going to guess at them.

On the capillaries, the plan proposes building new public express stations or converting existing facilities into them, and completing a county, township and village three-tier delivery system, with the stated aim of making it easier for people to receive online purchases and more efficient to move fresh produce out of the countryside. If your business has any rural collection or last-mile distribution in it, this is the heading that will eventually generate a local facility network you can plug into. If it does not, this heading is background, and it is perfectly legitimate to file it as background rather than as something to act on this year.

The 1,000 nodes figure needs one word of care, and the word is the verb. The plan says around 1,000 major freight multimodal nodes will be upgraded and improved during the 15th Plan period. Upgraded, not built. That is a programme about existing facilities getting better, and it matters for expectations: a node on this list is a place that is about to have money and engineering spent on it, not a new address appearing on the map. The first 300 are already being accelerated. If I divide 1,000 by the five years of the 15th Plan period I get roughly 200 a year, which is my arithmetic on a published count and a published period, and it is useful only as a sense of pace. It tells you this arrives steadily rather than all at once, and that being early in the queue is worth something.

Then the two growth figures, which are the ones most likely to be misused. In the first seven months of this year, national railway container rail-water intermodal volume rose 9.2 percent year on year, and rail-road intermodal volume rose 80.9 percent. Both are year-on-year growth rates over a seven-month window. A rate of 80.9 percent tells you the volume grew fast; it does not tell you the volume is large, and the two are not the same claim. Before anyone reads 80.9 percent as evidence that a mode has arrived, they need the absolute tonnage or container count behind it, and I do not have it. What the pair of numbers does support is a narrower and still useful conclusion: the intermodal channel is growing, and it is growing faster than the rail-water channel on these figures.

Now, the applicability question, which is where this column earns its keep. Before any part of this plan produces a consequence for your business, several things have to be true. Your lanes have to run through a corridor, a hub or a node that actually appears in the implementing documents, and those documents have not been published in anything I have read. Your mode mix has to be movable, meaning the freight you run is physically capable of the intermodal pattern the node is built for. And the follow-on instrument has to exist, because a plan does not disburse anything; an application notice does. Three conditions, none of them satisfied by the plan itself, which is why the honest answer to does this apply to me is: not yet, and here is what to watch.

Let me do the arithmetic, with every assumption written down. First the denominator trap, made concrete. Assume a company with annual revenue of 500 million yuan and annual logistics spend of 20 million yuan. Its own logistics cost ratio is 4 percent. If somebody in that company reads 13.1 percent and concludes the business is running nine points better than the national target, they have compared a company ratio to an economy ratio. The correct statement is that the company ratio is 4 percent of revenue and the national target is 13.1 percent of GDP, and the two are not commensurable. That is the whole finding, and it takes one sentence to state.

Now a transmission calculation, which is where the assumptions pile up and I want them all visible. Assume the implied baseline of 13.9 percent that I derived above. Assume the transport component of this company's logistics spend is 60 percent of the 20 million, which is 12 million. Assume that of the economy-wide improvement, half is transmitted to this company's transport costs; I have no basis for 50 percent beyond it being a round midpoint, and I say so. On those assumptions the relative improvement is 0.8 divided by 13.9, about 5.76 percent; the transmission to this company's transport costs is half of that, about 2.88 percent; and the effect on total logistics spend is 60 percent of 2.88 percent, about 1.73 percent.

Applied to 20 million yuan, that is about 346,000 yuan a year. I am not offering that as a forecast. I am offering it as a chain of four stated assumptions, any one of which you can replace with your own number and get a different answer. If you can only replace one, replace the transmission rate, because it is the weakest link and the one nobody ever writes down.

One more, on time rather than money, because intermodal is where the time cost hides. Assume a lane on which you have committed a 72-hour door-to-door service, and assume your current all-road transit measures 58 hours, leaving 14 hours of buffer. Assume that moving this lane to a rail-road or rail-water pattern adds two short-haul legs and one transfer, and assume the transfer plus waiting costs 10 hours; that 10 hours is my assumption and nothing in the plan supports it. Transit becomes 68 hours and the buffer falls to 4 hours, and the commitment still holds. Now assume the same lane actually measures 65 hours today, leaving 7 hours. Transit becomes 75 hours and you are 3 hours past your commitment. Same plan, same lane, same assumed transfer cost, two different answers, and the only thing that separated them was whether anyone had measured the current transit. That is the argument for measuring before switching, and it is an argument about evidence rather than about mode.

On the four things you manage, and keeping the time scale honest. On cost, the effect is not a price cut arriving in a quarter; it is a corridor cost structure that changes over four years as segments open and nodes are upgraded, which means your lane cost benchmarks need re-basing on an annual cycle rather than being locked in a contract for five years without a review clause. On lead time, corridor capacity expansion and smart upgrading are programmes that change transit reliability before they change transit speed, and reliability is the thing your safety stock is sized against.

On compliance, the exposure that this plan creates is documentary: if you intend to apply for anything under it, you will be asked to prove volumes, modes and facility use with records, and the burden of that proof sits with the applicant. On inventory, a denser facility network changes where you can hold stock, which is an argument for treating warehouse location as a decision that gets revisited against the node programme rather than as a fixed footprint.

On evidence, one point worth stating plainly. In an application-driven policy, the applicant carries the burden. Nobody is going to come and measure your volumes for you and then offer you something. If the record does not exist, the position does not exist, and the time to discover that is not three weeks before a notice closes. There is also a symmetry in it that I find useful: the same records that support an application are the records that let you evaluate whether the corridor change actually delivered anything, which means the cost of building them is charged against two benefits rather than one.

So, the records, and I will keep them concrete. A volume ledger broken down by mode, by lane and by month, kept continuously rather than reconstructed when something is due. A full-chain intermodal document set for every intermodal shipment, showing that the legs connect to each other and to one consignment, because a claim about intermodal volume that cannot be traced shipment by shipment is a claim about a total, not about a chain. An energy and emissions ledger for the equipment you operate, since the digital and green heading is where facility and equipment questions will land.

An asset and retrofit ledger recording what equipment you own and what you have spent on it, with dates, since retrofit is a fact about a machine and not about a fleet. A measured transit record by lane, recording actual door-to-door hours rather than contracted hours, because every decision in this plan turns on the difference between the two. And a correspondence and submission log recording what you filed, when, and what came back, with the reference number.

On the calendar, on a quarter and year scale because that is the scale the plan runs on. By the end of September 2026, the planning function should have mapped the company's top lanes against the four artery programmes named in the plan and marked which ones are plausibly affected; I would set the threshold at the top twenty lanes by spend, because below that the mapping costs more than it returns. By the end of October 2026, the same function should have identified which provinces or municipalities relevant to those lanes have issued implementing documents, and should be tracking those publications on a monthly cadence.

Before the end of 2026, the volume ledger and the measured transit record should be running, with at least one full quarter of data in them, so that anything you file in 2027 rests on measured rather than estimated figures. Through 2027 and onwards to 2030, review the lane map annually against the node programme, with the first 300 nodes as the near-term watch list and the remaining 700 as the medium-term one, and re-base lane cost benchmarks once a year rather than at contract renewal.

On alternatives, and what each one costs. Doing nothing is a genuine option here and should be named as one, because this plan does not oblige a private company to do anything; the cost of doing nothing is that you find out about corridor changes and funding windows after the queue has formed, which is a cost measured in position rather than money. Joining a hub or node programme early buys position and costs management attention plus whatever documentation the application demands, and the risk is committing facility or equipment spend against a node that slips, since a five-year programme of around 1,000 nodes will not deliver on a uniform schedule.

Re-basing contracts annually to allow corridor re-pricing costs negotiating effort and protects you against being locked into a lane price that the network has moved away from. Waiting for the implementing documents before building any records is the option I would argue against, and not because of the policy: it is because the ledger you build now is the only way to know later whether the change did anything.

And the pitfalls, which are mostly versions of the same mistake. Treating 13.1 percent as a company target, when it is an economy ratio. Treating an implementation plan as a subsidy programme, when the money arrives through separate instruments with their own application requirements. Treating around 1,000 nodes as 1,000 new facilities, when the verb is upgrade and improve. Treating 80.9 percent growth as evidence of scale, when it is evidence of rate and the base is not published in what I have read. Treating the seven headings as equally relevant, when the policy support heading is the one that converts the others into something actionable. Assuming a corridor programme covers your lane, when the segment lists are in documents I have not seen. And assuming that intermodal saves money by definition, when it saves money on some lanes and on some lane lengths, and the only way to know which is to measure the transfer cost before you commit.

Here is the exception I want to leave you with, because it is the one people get wrong in both directions. Nothing in this plan requires a private company to do anything at all. That is true and it is worth saying plainly. It is equally true that the network this plan builds is the network your freight will run on in 2030, and that the companies which will be well positioned on it are the ones that measured their lanes and built their ledgers while there was nothing at stake. Neither of those statements contradicts the other, and holding both at once is the reading I would defend.

This is a four-year document with a five-year facility programme inside it and a target year of 2030. It will not produce anything you can feel this quarter. What it produces this quarter is a decision about whether you start measuring now, and that decision is cheap, reversible, and the only one on this list that does not depend on a document nobody has published yet.

↑ Back to the key points

— By Dr. Ingrid Voss