In an earlier session, we saw the drivers of flow rate. We learned that the flow rate was driven by the minimum of demand and capacity, where capacity in turn was driven by the bottleneck of the process. In this session, we'll discuss the drivers of inventory. Why should there be inventory in the process? Would we all want to be lean and live without inventory? We'll discuss a couple of reasons why firms might hold inventory. We'll see that variability in particular is going to be a big driver of why firms hold so much inventory. Now, keep in mind as we're doing this that everything that explains inventory will implicitly also explain the flow time. We know we're a Little's law, holding the flow rate constant, everything that we do to inventory we will do to flow time. So while we call this session, five reasons for inventory, we could really also call it five reasons for longer flow time. In our calculations for the Subway case, we assume that workers would work like robots. It will take for example a worker exactly nine seconds per customer to put the cheese on the sandwich. We will stick with this assumption for a little longer, but at this time I want you to understand the implications of assuming that these processing times have absolutely no variability associated with them. Imagine you're working in a food truck. Say for the sake of argument, that at any time slot of five minutes duration you can get between 0, 1 or 2 orders with equal probability. Also, have a capacity of 0, 1 or 2 orders. Now it's also not possible to make a sandwich before the orders are occurring and customers are not willing to wait for the sandwich. How many sandwiches will you sell every five minutes? Well previously we defined the floor rate as the minimum between demand and capacity. In this case, since we have an average demand of one, and an average capacity of one, a naive analysis will suggest that our average flow rate is one. However, this example is misleading. Let's look at all the things that can happen at the food truck. We can have a demand that is either 0, 1, or 2. We can also have a capacity that is either 0, 1, or 2. That leads to nine cases. If we have a demand of 0 and a capacity of 0 well obviously the flow rate is 0. The flow rate is also 0 for a capacity of 1 or 2, as long as the demand is 0. Remember, the flow rate is the minimum of demand and capacity. So I apply the same calculations, the minimum between demand and capacity, to these nine rows here. I noticed that the average flow rate is much lower than I initially expected. Remember each of these rows here is occurring with the same probability. And I see that the average flow rate that I obtained this way is just 5 over 9. Put differently, I've lost almost half of my flow rate just because of the effect of variability. Now people refer to this effect as buffer or suffer. Buffer or suffer means that in a system that is not able to buffer the variability we will suffer a loss of flow rate. Remember our crucial assumption here. The food truck is not able to make a sandwich before the order occurs, and customers are not willing to wait for it. So we explicitly eliminate inventory. Both inventories of waiting customers and inventories of waiting sandwiches. By allowing for inventories, we'll actually be able to have a higher flow rate. If we allow the food truck to make a sandwich to inventory is a situation where the demand is 0 or vice versa. If we allow the food truck to inventory to buffer some customers when the demand is high, the flow rate will go up. So in other words, inventory helps us with the flow rate by decoupling the supply and the demand. This is one of the reasons why companies like to have inventory. Buffer or suffer, there is no right answer to the simple question. You notice this by comparing the two biggest restaurant chains in the world, Subway and McDonald's. At Subway, as we already discussed, customers wait for their sandwiches to be made because the sandwiches are made to the customer order. The customer orders the sandwich and then stands there in line potentially waiting with other customers, until their specific sandwich, uniquely made for them, is completed. McDonald's follows a different strategy. At McDonald's the restaurant makes a whole batch of sandwiches. They don't make just one cheeseburger, they just do twenty in a row. Then the sandwiches will wait for the customer orders. So instead of the customer waiting for the sandwich, at McDonald's, you have the sandwich waiting for the customers. The benefit of that is that when the customer order finally comes in you don't have to make the sandwich, you can fulfill it immediately. So instead of having the customer wait for sandwiches, the sandwich waits for the customer. We will call the McDonald's approach, make to stock, or make to inventory. The main advantage of make to stock is that it allows for scale economies in production. You going to produce 20 hamburgers in a row. The second advantage of it is it also allows for a short flow time for the customer order. Now notice that you have a longer flow time for the hamburger, but a short flow time for the customer order. In the Subway setting, the benefits of make to order are that you can make the sandwiches fresh. The flow time for the sandwich are short, which comes at the expense of the waiting time for the customer. It also allows for more customization. It would be impractical to hold all potential versions of a Subway sandwich in inventory. By making the sandwiches to order, you can allow the customer to choose whether they want extra mayonnaise or no onions on their sandwich. The final benefit of made to order, is that your producing in exactly the quantity of demand. We'll revisit the question of make to order, make to stock, when we talk about customer choice. Because make to order allows you, as we discussed, to dramatically increase the level of customization in the production. Though inventory costs us money to store as we've seen in the calculations of inventory turns, the examples of McDonald's have shown us that management might sometimes be well advised to hold inventory. Let me propose five reasons why management might want to hold inventory in the process. There might be other reasons but I think these are the most important ones. The first reason for inventory is what I call pipeline inventory. Pipeline inventory is a direct result of Little's Law, I =R *T. If you are producing wine and that wine needs to age for two years before you sell it, well you need to hold two years of wine sales in inventory. The second reason for inventory is called seasonal inventory. Imagine you're a toy manufacturer and 90% of your sales are happening around Christmas. With that in mind, given that you want to produce at a relatively steady pace all over the year ,very likely in October and November at the latest, you're building up very significant inventory. The third reason for inventories called cycle inventory. Oftentimes as we already saw in the McDonald's example, economies of scale in the production of purchasing. Just think about how in your household you are purchasing beer, wine, milk, orange juice or whatever your favorite beverage is. You might be drinking a bottle of beer a day. However it's unlikely that you're going to run to the beer store for every individual bottle. You're buying a case and then the other bottles sit there in inventory. We've already seen in the McDonald's example a case of safety inventory. McDonald's doesn't want to run out of hamburgers to sell. So they're producing the burgers ahead of demand. They always keep a little bit of safety stock, just to make sure that their customers don't have to wait for the food. Finally, the last reason for inventory is a decoupling inventory. Instead of just buffering us to external demand from the customers, wemight also want to buffer us from internal demand from other process steps in the operation. So the last three examples here are really driven by the variability in the spirit of buffer or suffer. The previous two are driven really by variations in the supply rate to match a seasonal demand. And the first one, the pipeline inventory, is just the minimum inventory any operation needs to exist. Buffer or suffer. Inventory can be used to shield us from the variability of demand. Inventory can also be used to shield us from internal variability, be it in the form of setup times or variability in processing times. For this reason, inventory is really a very powerful symptom that guides you in proving your processes. Wherever you see inventory, you will have supply and demand mismatches. And those tend to be the biggest opportunities for process improvement. In this session we also talked about the difference between make to order and make to stock. By following make to stock model a firm holds inventory even if it has not yet seen the demand. This allows the operations to look fast to the customer. Remember, the customer doesn't want to wait. The operation looks fast to the customer, even though internally it might actually be very slow.
