After returning from my mother’s 100th birthday celebration, a few siblings and I started reviewing our savings to determine if we would have enough to live our lifestyles were we to live as long as my mother. A couple years ago, a financial planner had told me, “Your funds will last much longer than your 100th birthday.” I’d done the spreadsheet budgeting and used on-line tools, which all said what the planner said. Entering in inflation, investment return, monthly expenses, my retirement savings should last to leave my grandchildren college funding.
Then, I got a call from my Schwab advisor who does an annual checkup. She told me that using the same numbers, I have a 40% chance of ending up living in my car. After I calmed down, she explained. The various calculators and my spreadsheet all share a simplification that makes them blind. It’s the same kind of simplifications that are buried in the county’s water modeling and forecast. When you enter in your assumption of a 3% return on your investment, the calculators make 3% the return for every year. In reality, 3% might be the average, but can it swing around from large negative numbers to nice high ones. There is uncertainty in the return, even if 3% is the long-term average. If the first year of the plan sees a down market, say 20%, which is possible, then an average of 3% thereafter will not come close to the goal, even if year 2 is +20%. There is randomness and uncertainty that are not considered in simple forecasting tools. The uncertainty that can make a huge difference, and often does.
The Schwab advisor explained that they run Monte Carlo simulations, in which each year has some uncertainty in it, with the uncertainty taken from the range of possible values. They don’t plug in 3% for each year, but roll the dice, and plug in the value that comes up. When they have finished, they do it again, the same way. The dice may come up different. They do the exercise 100 times or so, and then show how many of these runs come up with hitting the target or better, and how many fail.
The Monte Carlo method is well-known and used a lot in planning and forecasting; it’s important because there is a lot of uncertainty in the world, and more uncertainty ahead. We want the risks to show up when we do these exercises, so we can plan to avoid them as much as possible.
How is this related to our water planning in Napa County?
The large enterprise the county is engaging in to comply with the State Groundwater Management Act (SGMA) includes a very complex and expensive modeling of the groundwater and the inputs and outputs to the water going into and stored under the valley. Only a few trained and certified technical people can understand and use these modeling programs; that’s how complex and arcane they are. One thing is known, they do just like the spreadsheet does. They simplify by using a fixed value for each cycle instead of doing Monte Carlo simulations to include uncertainty. They could be wrong by a significant amount, due to this simplification. What’s more, when one uncertain value is in a computation with another, the results can be wild and unexpected, just like life is.
Stripping out uncertainty in forecasting is stripping out the very thing we are trying to plan for. We use fancy computer tools to show us what we might not be able to see with simple logic or intuition. We use modeling to surface scenarios that didn’t occur to us, or ones we thought unlikely until the data show us differently.
The useful output of a modeling exercise is the emergence of the unexpected, the results that make us say, “Oh, S@#@! I hadn’t thought of that” When the output is, “That makes sense; no big surprise”, we might not be using the right method or maybe we aren’t spending our time and money well.
We treat many things as constants that are actually uncertain, and getting much more uncertain in climate change. The seasonal rainfall, the concentration of the rainfall, the temperatures, fires, the availability of State Water Project supply, and other factors can combine in ways we aren’t fully aware of. The purpose of using modeling is to make us aware of how they can combine in ways we can’t easily foresee. Without proper modeling of uncertainty, we can easily comfort ourselves that everything looks good because we spent a lot of money on consultants who tell us that things look good.
On the happy occasion of my mother’s 100th, a couple of us were triggered into doing a planning exercise that showed the importance of including uncertainty into our plans. Those plans are simple compared to forecasting our water supplies. On the stressful event of accelerating climate change, our water planners should be studying how factoring in uncertainty can make us less blind in our planning.
We should not be comforted by volumes of meetings, reports, and moneys spent unless methods and tools are updated to reflect our new reality.