Milind Sohini did his BE (Mech) from COEP. He worked with Thermax for a few years before leaving for UT Austin to do his Masters in Manufacturing. He then went on to do his PhD from Georgia Tech. He worked for a few years flying planes, ok make that scheduling planes, for Delta Airlines before jumping into academics. Although his doctoral program was at an engineering school, his employment has been mostly with B Schools. He has spent almost two decades at ISB, Hyderabad, before moving in 2023 to SUNY, Buffalo where he is the chair of the Department of Operations Research. I have met him a few times at ISB, and I have always wanted him to talk to our mechanical batch of1990 and take us through some of his work in supply chain. That finally happened last week.
What comes to your mind when you think of a supply chain? Efficiencies, Systems, Movements, Storages. To sum up, stuff that minimises costs. Milind started by questioning this cost drive perspective and the secondary role that supply chain usually plays in organisations. Even for companies that are primarily in logistics, supply chain still plays second fiddle to the dons in marketing and finance. Milind talked about an Indian e-commerce example. Flipkart’s Big Billion days was an idea driven by the marketing folks. Whereas supply chain if accountable for the present, marketing is accountable to the future. So how effective was the Big Billion day sales pitch? Did the discount conscious customer buy any non-loss leader products on those days? Did these customers stick around and demonstrate loyalty to the platform once the sales days were over? We are not sure. What we are sure of is that handling these sudden surges in sales was a supply chain nightmare. Lots of unshipped orders. Lots of mis-shipped orders. Huge costs of reverse logistics.
Why do such things continue to happen across companies and across time? The simple answer lies in the way incentives are designed. Sales folks are incentivised on revenues, logistics folks are incentivised on costs. So each department drives decision making based on its own incentive design. Even on the supply side, production will decide to lower costs by making everything in batches, long production runs. Logistics folks also want to supply in batches, say full truckloads. But very rarely do these both batches match. The result – build up of inventory.
Milind took us through a live example of how wrong incentive designs can create huge management challenges. He took an example of the public healthcare system. In developing countries like India, one of the significant challenges comes from managing diseases like Tuberculosis (TB). Milind spent some time in Public Health Centers in Dalsinghsarai, Bihar, understanding this problem. 40% of Indians have the TB bacillus residing inside their bodies, waiting for episodes of reduced immunities to come out to attack. The most affected class of people are those involved in physical labour – your typical daily wage earner.
Treatment for TB is a long drawn process – typically taking 6 months for a patient to recover with the streptomycin treatment. The first two months are the most arduous – the patient is so weak from the antibiotics that he/she is almost bed-ridden. For a daily wager, this period is a total loss of income. Patient starts regaining strength after this – and in the belief that things are normal now, they stop taking the antibiotic. The bacillus bounces back – and it will take an even higher potency antibiotic to handle the disease – stretching recovery from 6 months to 9 months.
The government has an interesting scheme for ensuring compliance in regularity of taking medicine. They actually send ASHA workers to the patient’s home with the medicine. The ASHA worker also gives a glass of milk and an egg along with this medicine. Side effect: the neighbours have also started reporting TB cases in their families.
The PHC is not really a place that patients go to first when they get TB. The first responder is a Rural Medical Practitioner (RMP). A typical RMP has worked as a doctor’s assistant. In my younger days, we would call such people compounders, as they would, in the absence of pharmacists, literally create compounds of drugs based on the diagnosis the doctor would make. These folks operate at the bottom of the pyramid – and charge just Rs. 20-30 for a consultation. In healthcare, practitioners love repeat customers – as the incentive is based on disease and not cure. So the RMP starts off the TB patient on crocin, even when he suspects TB. The crocin treatment continues for two weeks – and the RMP has earned about Rs. 200 – Rs. 250 from this patient during this time. Although the RMP has the interest of the patient in mind, the perverse incentive of disease vs ease, makes him delay referring the patient to a bigger city like Patna. This two week delay also means that the antibiotic treatment potency and costs are both going to go up.
The Bihar government looked at the RMP incentive structure and decided to incentivise the RMP by paying him Rs. 250 for every positive TB patient that he refers to the PHC system. This incentive led to some very interesting behavior. The TB bacillus resides in the lungs. So to test the patient for TB, you need sputum that has come from the lung. If the patient is tested during the initial days of infection, the test very often comes negative, as the bacillus has yet to reach the lung. So for the RMP to earn his money, he has to let the disease progress for some time – which in turn is not really good for the patient’s health and wealth.
Lest you believe all is lost in health, Milind went on to talk about how technology has managed to improve delivery in public healthcare systems. For most vaccines to be effective they have to be kept below a certain temperature from the time that they are manufactured to the time that they are injected. If there is any break in this cold chain, the vaccine loses effectiveness. So this is yet another instance of the supply chain not being just driven by costs, but by the very efficacy of the product. In Indian villages, vaccines are kept in kirana stores next to Coke bottles. Not only is there uncertainty of power supply for the refrigerator, but very often in summers, the retailer will keep the vaccine out to accommodate higher revenue generating Coke bottles. What is the solution? What is important is to have transparency in the entire supply chain. Some kind of block chain where temperatures are tracked for a consignment all the way from the Serum Institute factory to Dalsinghsarai. And a SOP to provide corrective action when it is detected that the cold chain is about to be broken.
Some of us were at sea when Milind mentioned jargon like block chain, so Milind took us through a crasher on the same. A blockchain is basically a list of transactions. The idea is that instead of a central repository of data, we use crowd storage. Some of these transactions can be financial ones, say for example, in cryptocurrencies. So you need cryptographic tools to ensure that only those with the right keys can decode the information in these transactions. In the democratisation of data storage, there is a high chance that dishonest players will not want the complete records to be shared. The algo used in such cases is to choose the longest blockchain, as it will probably have the complete record. Blockchains are being used by DeBeers to map the supply chain for diamonds. Academic institutes like MIT and iSB share student transcripts using blockchain.
From blockchains we meandered to AI/ML. Data is information when it can lead to action. With IoT devices becoming ubiquitous, we have a lot more data available now. (Fun fact pointed out by Vivek who has spent a long time with TCS – no data also is data, meaning that probably your sensor has conked off 🙂 ) What has also changed is the way the data is being gathered. Earlier most data coming into the system was text, but with better bandwidth it is increasing images and video. Indian railways has started deploying cameras that scan the undercarriage of coaches to decide repair tasks even before the rake reaches the maintenance depot. In a typical factory setting, motors are sending servers real time data about vibration, current and temperature. Information for the factory is knowing when to replace or service a motor.
Data is also helping transform product companies to service companies. Service organisations are characterised by production and consumption happening at the same time. Milind went back to the airlines sector to make this point through an example of Rolls Royce, the pioneer in jet engines. In the 1950s when jets first started being made, the typical life of an engine was 8 years. With improved design and manufacturing, it has increased to 25 years today. This has not made Rolls Royce folks very happy – as they have been losing the engine replacement business. Somewhere in the 1990s, Rolls Royce converted its engine product into a service. Airlines are now sold engine uptime instead of the engine itself. Power, by the hour. Usage not ownership. Sensors collect data at the end of every flight. This data is analysed across planes, across airlines to arrive at optimal preventive maintenance schedules. The engine manufacturer can do a much better job of this than the airlines themselves.
But supply chains are not just about technology. Tech companies use old fashioned monopoly tactics to generate supernormal profits . Let’s take an example of the world’s most profitable company, Apple.We assume that Apple has conveniently delegated all manufacturing and supply chain management to the likes of Foxconn, but nothing can be farther from the truth. For any company, make or buy is a very big decision. Apple’s pricing power depends on their control of the supply chain. Every six months, the Apple design team comes up with a product innovation that requires a relook at manufacturing. To cite an example, when the iPhone was launched, it had an aluminium back cover which required precision milling . Apple went on to contract the global supply of CNC milling machines – in effect, choking the market. Competitors who wanted to copy the product, had to wait 6 months for the capacity to come back on stream, by which time Apple was already on the way to introduce its next innovation.
Not all of us are blessed with Apple’s deep pockets and huge cash reserves. If you are a competitor to a large player like Apple, what is the best way to tackle Big Brother? Milind talked of using information asymmetry and how small players can profit from it. He went back to his favourite airline sector to drive home the point. The revenue from a seat is lost when the plane doors close. Most airline websites will not disclose how many seats are booked and how many are empty. Code sharing is when a local airline ties up with an international airline to offer its customer a turnkey solution to her entire journey. This is an easy way to expand networks and fill up empty seats for both the airlines. For example, Delta Airlines partners with Vistara so that a customer can book a flight from Pune to Detroit. The Pune Delhi sector is taken care of by Vistara and the Delhi New York and New York Detroit sectors are taken care of by Delta. Usually airlines share revenue on a mileage pro rata basis. If Vistara does not disclose where the passenger is boarding the flight in India, it may be able to command a higher share of revenue.
About 13 to 14% of India’s GDP is tied to logistics and supply chains. (It is around 7% for developed nations.) The conventional cost minimisation model has now been challenged by new age logistics companies like e-commerce giant Amazon. We need to learn from the supply (and demand) chain to not just reduce costs. A change in mindset is the need of the hour. A supply chain focussed organisation where folks gets incentivised on profits, instead of sales or costs.
Additional Notes – Not to be included in Blog
Any cost driven approach relies on prices for signalling efficiencies. This assumes perfect availability of information, which rarely exists in real life markets. In fact, information asymmetry is the basis for any arbitrage. Is cost minimisation the best way to optimise a supply chain? Doesn’t seem to be the case. Let’s look at the new buzz word for global businesses – Sustainability. It adds costs – which finally is being paid for by the end customer.
If you are a company which can charge a premium for its services, costs may be the last thing on your mind.
We digressed to quality – and how it needs to be driven through the supply chain. Omkar Kendhe talked about his experience with the auto industry. About a decade ago, many auto OEMs realised that they don’t have the bandwidth to manage relationships with too many vendors. They shortlisted the larger vendors – let’s call them tier 1 vendors – and decided that the remaining vendors will now become suppliers to these vendors – tier 2. On the same lines you had tier 3 and 4 vendors. The moot question in this system: Does a Tier 1 vendor have the wherewithal to train tier 1 vendors and help them improve quality?
DuPont is one of the biggest players in the GMO seeds business. The Indian seed market has a very short sales window. Most sales happen in June, around the time the monsoon starts. There can be huge variability as this is also the time that the government announces its subsidies and MSP for different crops. Seeds are living products and have limited shelf life. If they are not consumed in that cropping season, then the mandate is to destroy them. To insulate the retail chain against the vagaries of demand, companies like DuPont has a liberal buyback policy. The company . takes back the entire stick of unsold seeds and pays the retailer the same price that he bought it for. DuPont has enough margins and reserves to handle this stock. Also being present in multiple countries, it can move around this seed stock to sell it before it expires. Some of their contract farmers had been pilfering seed and selling it in the gray market for almost an 80% discount on DuPoint’s pricing. Milind did work with DuPont – and his advice for managing this problem was simple. Rely on pull more than push for seed sales. If the farmers ask for DuPont seeds then the gray market seed will not create problems.
Coopetition is when competitors come together for a supply chain, as it benefits both to work together. Milind cited a case about the Canadian railroad sector to illustrate this point. Canadian Pacific railway is headquartered in Alberta, Western Canada, and sources most of its customers from that area. Canadian National Railway is headquartered in Montreal, which is in Eastern Canada. Both operate single track routes from East to West Canada. Working together and dedicating lines to one direction makes sense for both players as it helps rake utilisation for both parties.
Next, he took us through a case where there are two competitors who are buying components from another city – but not in enough quantities. Full truck loads work out cheaper – so both decide to get together to transport using a single truck. Say one player has a 75% share of the market and the other has 25%. Now that both have reduced their costs, there is a chance that the smaller player, taking advantage of the lower costs, reduces prices to gain market share. In which case, coopetition can backfire. In such a case it is better to raise rival’s costs. (In cases where both parties are equally sized, coopetition might work better.) Milind recommends that you can read some of the stuff on coopetition by Avinash Dixit, Prof at Princeton University.
You have a supplier who is capacity constrained – and uses a strategy of rationing out his production. So when you order 100 cases, you end up getting only 50. So you game the system by ordering 200 cases, knowing that you would get 100 that way. In such a case, the supplier does not want you to know what capacity he has.
Take the example of an industry with a very fragmented supply chain – textiles and garments. Not only are we looking at the energy and water that goes into manufacturing a T shirt, but we also need to be cognizant of the water (and detergent) that gets used for the 30-40 wash cycles that a typical T shirt goes through in its existence. Poorer countries suffer higher ecological damage, as governments there prioritise consumption over pollution. A good first step, minimise total costs.