Thinking about Late Policy

~1200 words, ~6 min reading time

I’m thinking of revising my late policy. For philosophical reasons, I don’t penalize grades simply because assignments are late. Instead of grade penalties, I require students to email me a late form which explains why the assignment was late, and has them say what they could have done to prevent it from being late. The idea is: (1) to make it a teaching moment, so students are more prepared when they face real, serious deadlines(“Hey, maybe I shouldn’t have waited until 11:55PM to submit something due at 11:59PM, because sometimes the internet goes down.”), and (2) to let the punishment fit the crime. Grading a late assignment is mildly inconvenient for me, so the punishment should be similarly mildly inconvenient.
However, this semester I noticed something: there are two different types of late assignment doers.

Type I (the most common): people who hand things in a little bit late, usually because of poor foresight. (Took longer than expected, computer problems, forgot about the assignment but had to work that night, etc.) Generally speaking, these students hand in very few assignments late. They intend to hand things in on time, and have reasonable (but imperfect) systems for getting things done on time.

Type II (the less common): people who hand in a bunch of stuff at (or near) the very end of the semester. This seems to happen for two reasons: (1) a strategic decision to prioritize work in classes with less forgiving late policies, and/or (2) some significant life event which then leads to poor mental health which creates a more severe interference with executive function. (The life event might just be poor mental health itself.)

It makes little sense to treat these two the same, as they are very different problems. The Type I case isn’t really a “problem” in any significant sense. These students just need to learn to give themselves more of a buffer than they think they need. Also, the work created for me from these students is pretty minimal, both individually and collectively. If you turn in an assignment before I even grade those handed in on time, then it’s really no less convenient for it to be late than for it to be on time. The late form method was designed with this type of case in mind, and I think it works well for this purpose.

The Type II case is a bigger problem from a learning perspective, as the “do everything at the end of the semester” method has been shown to be bad for academic performance and bad for students’ mental health. (So, this is especially bad if students are doing it because their mental health is already poor!) Plus, more selfishly, I find this kind of behavior extremely annoying, as it *does* increase my workload at a peak time in the semester, and the wide variety of assessments I’m being asked to grade (going all the way back to week 1!) requires a lot of mental switching which reduces grading efficiency. I don’t think late forms are the right intervention here. Instead, I think a heavier-handed approach is appropriate. Specifically, I think it would be good to meet with these students individually to talk about their situation. (This also has the added benefit that I am much nicer in person than via email, as it reminds me that I’m dealing with actual people.)

A simple method I’m thinking about using:

(1) If an assignment is < 1 week late, only the late form is required.

(2) If an assignment is > 1 week late, you have to meet with me to chat before I’ll accept it.

The problem I foresee with this is that I’d be increasing my end-of-semester workload significantly, as I’m adding meeting with students in addition to grading all the late work, giving finals, etc. However, there are two factors offsetting this concern: (1) the pattern seems to be a few students (<5%) turning in a large volume of end-of-semester late work, so I really wouldn’t be meeting with very many students, and (2) meeting with me is a bigger hassle for students, so that would encourage students to turn things in closer to on time.

But, to further offset this concern, what I really need is an *early identification* system so that I can intervene before things pile up too much. With my grading system, this is tricky. The system is designed so that students can choose their goal grade, and then have some choice about how to achieve this grade. So, when Student X doesn’t do Assignment A, it doesn’t necessarily mean much. They might just not need it for whatever their goal grade is.

I tried to provide some guidance this semester by giving students occasional “Grade Updates”, where I told them what work I had received and graded and what work they still needed to do for various grades. However, a few students still fell through the cracks here. (Some of whom just weren’t in class to get the Grade Updates, and some of whom either misunderstood the grade updates or just ignored them.) I spent quite a bit of time saying to myself “This small group of students isn’t going to pass the class. They come to class, but don’t do the work (which is almost all on Blackboard), and I can’t figure out WHY.” (Of course, I didn’t directly ask them. Apparently, I prefer just being baffled by this kind of nonsensical student behavior to gathering data. That’s the theorist in me, I suppose.)

Now, I could try to intuit my way to a system that identifies students that need more active intervention – but that seems likely to be unbalanced and open to my own biases. (For example, I have a preference for students that sit close to the front and participate in discussions. Students that hide in the back or don’t come to class are much less likely to catch my attention.)

A couple of options:

(1) Use statistics to figure out which of the early assignments are most strongly correlated to failing the course. This would allow me to flag those students that don’t do these assignments. This also saves me having to check in with students continuously – just focus on a few key assignments to figure out who to intervene with. Downside: correlations aren’t perfect, so still plenty of cracks for people to slip through.

(2) Have some progress standard triggers at various points in the semester. This minimizes the number of students that would fall through the cracks. Downside: I might end up triggering lots of meetings, which could be very time-consuming.

It’s a tricky thing figuring out how to strike the balance between giving students their independence (which is the side I tend to err on) and providing sufficient guidance for those students that need more extrinsic motivation in order to succeed. But, I suspect that, as our society is still going through and then beginning recovery from the pandemic, a more active role may be useful.

December Inflation Update

~600 words, ~3 min reading time

On Friday, the latest CPI numbers came out. We’re looking at a 6.8% increase in prices from November 2020 to November 2021, and a 0.8% increase from October 2021 to November 2021 alone. (In annualized terms, that’s nearly 10%.)

By Usonian standards, this is quite high – literally the highest rate we’ve seen in my lifetime. (Last time we saw an inflation rate this high was about 9 months before I was born. Coincidence? Almost certainly!)

As before, this rate is largely reflecting rising gasoline and fuel oil prices and rising prices for vehicles (both new and – even moreso – used). However, core inflation (which excludes energy and food, since those prices tend to be very volatile) is up 4.9%, which is a bit disturbing.

Also a bit disturbing is that the U of Michigan survey is showing inflation expectations of about 5% over the next 12 months. The 5 year TIPS breakeven inflation rate shows an expectation of averaging about 3% inflation over the next 5 years – about 0.5% higher than a month ago. Now, the Fed says they’re looking to target “average” inflation of 2% – though are *very* vague about over what time frame.

If we’re being forward looking, then literally every TIPS breakeven inflation rate is showing expected inflation over 2% – whether you’re looking at a 5, 7, 10, 20, or 30 year span. (This has been the case since early 2021.)

Looking backward, the ONLY time frame where average annual CPI inflation has been under 2%, ending in November 2021, is if we choose 2008 as our starting point.

All to say, it would be very hard for the Fed to justify *not* tightening up its policy at this point.

UPDATE in response to a question on Facebook. The question:
We were worried about a lack of inflation over the previous 12 years or so right? I remember headlines ” Why us inflation so low?” Is this partially an adjustment combined with supply chain problems and a pandemic?

My response:

I think it depends who you mean by “we”. But, there were some concerns for a while, though from October 2016-January 2020, CPI inflation was hovering consistently between 1.5% and 3%. So, it feels to me like that problem had faded.

Lots of things are going on that could explain this. Nominal GDP is basically back on trend now. Real GDP is a little higher than where it was pre-pandemic, but definitely not back on trend. Put another way: spending has recovered, but production hasn’t yet (running some quick stats, I estimate we’re about 2-3% below trend) – so we’re spending that money by paying higher prices rather than on (much) more stuff. This is consistent with there being some lingering supply-side issues.

On the monetary side, we saw a big increase in M2 money supply in early 2020, but M2 velocity tanked at the same time – people were basically just holding the new money rather than spending it (not surprising given the combination of virus fear and lockdowns). Since then, the money supply has continued to increase (though at a somewhat slower pace), but velocity has held constant – the new money is actually being spent basically at the pace that it’s being created.

Now, there is something of a philosophical debate here in terms of what monetary policy should do. NGDP targeters are, I suspect, fairly content with things at the moment. (Though Scott Sumner did recently suggest inflation rates are too high, and that the Fed should focus on fighting inflation – but his estimates for ideal inflation aren’t way off of current levels.) Meanwhile, those that are more concerned about money growth, price levels, or inflation rates are more disturbed. (In July, John Taylor compared the Fed’s current stance to that of the Arthur Burns Fed of the 1970s.)

No Time to Lift? A Quick Summary & Exercise Routines

~ 1250 words, ~6 min. reading time

I ran across this recent paper by Iversen et al. Basically, they were trying to figure out how to make a time-optimized lifting routine for people who don’t have much time to get to the gym. They have pretty good summaries of their findings, so I’m going to offer a couple quick routines for people with little to no equipment.

Routine 1: “I have dumbbells in a variety of weights (or adjustable dumbbells)” 2x per week to progress, 1x per week to maintain

This is the kind of routine I would do if I was more pressed for time than I am.

Set 1: 10 dumbbell squats (Deload rule: if the weight you choose makes it so you can’t do these, decrease the weight before the next set)

Rest 2+ minutes

Set 2: dumbbell squats, as many reps as possible, up to 20. (Deload/Progression rule: <10, decrease weight 5 lb per side for next work out. 10-11, keep weight the same for next workout, 12+, increase weight 5 lb per side for next work out, 15+, make that 10lb.)

Set 3: 20 reps of dumbbell floor press using a “rest-pause” technique. Do as many reps as possible (track reps for the first set), then pause for 20 seconds. Then, do as many as possible, pause 20 seconds, and so on, until you have a total of at least 20 reps. (Deload/Progression rule: <10 reps in first set, decrease weight 2.5 lb per side for next work out. 10-11, keep weight the same for next workout, 12+, increase weight 2.5 lb per side for next work out, 15+, make that 5lb.)

Set 4: 20 reps of one-arm dumbbell rows using rest-pause, as above (do each side separately, and start and count first set with non-dominant side).

I suspect this would take about 15 minutes to complete, maybe less. And, if it’s more convenient for you, you can break this up over 3 days – so instead of working out 2 days a week, you work out 6 days a week, but the “workout” is a single exercise that can be completed in under 5 minutes.

What if you have more time? In that case, increase the number of sets of squats (2+ min rest between), always progressing/deloading based on the last set, and increase the number of reps of the other two exercises, progressing/deloading based on the first set of those.

Reasoning: evidence is that people can generally progress with as little as 4 sets per muscle group per week. Iversen et al suggest using 3 exercises: a leg exercise (like squat), a push (like floor press), and a pull (like rows). Now, technically this looks like 4 sets of legs and 2 sets of push and 2 of pull per week. HOWEVER, Iversen et al. also cite research that using rest-pause for 20 total reps is similar to doing 5 sets of 4 (and actually might be better by some measures). Also, training TO failure instead of just “close to” failure is way easier for those of us without as much bodily awareness. If you ask me “How many more reps could you do?” I literally don’t know. It’s just something I can’t feel easily. But, I *can* feel when my body just doesn’t want to do any more – like if I can get a partial rep done but no more. Why not use rest-pause for squats? Rest-pause can be pretty intense, so it is generally not advised for “big” exercises. I’m pushing it here with the floor presses because, in my opinion, dumbbell floor presses are much safer than barbell bench presses. The reason is simple: if you fail a dumbbell press, you can drop the dumbbells on the floor. But, if you fail on a barbell press, you can end up dropping the bar on your neck, causing serious problems – maybe even killing you. So, DON’T do rest-pause for barbell bench presses. Always be aware of what happens if things go wrong. In terms of rep choices – I recommend 10 because that’s enough that warm up sets are not necessary.

Routine 2: “I have me!” 2x per week to progress, 1x per week to maintain

Suppose you have no actual equipment. Perhaps you’re just starting or you’re on vacation. Now what? I highly recommend looking at Start Bodyweight. It turns out that your body doesn’t *care* where resistance comes from. Here, I’m just boiling down based on Iversen et al’s principles.

Exercise 1: 20 bodyweight squats using a rest-pause technique – do as many as you can (pick a variation where you can do 10-20 in the first set), rest 20 sec, then do as many as you can again, etc. until you get up to 20+ total.

Exercise 2: 20 pushups using a rest-pause technique – do as many as you can (pick a variation where you can do 10-20 in the first set), rest 20 sec, then do as many as you can again, etc. until you get up to 20+ total.

Exercise 3: 20 horizontal pulls using a rest-pause technique – do as many as you can (pick a variation where you can do 10-20 in the first set), rest 20 sec, then do as many as you can again, etc. until you get up to 20+ total.

Progression: try to add a rep to the 1st set each workout. Once you get up to 20 in a single set, move up to the next variation

What if you have more time? Increase the the number of reps, or add additional exercises from Start Bodyweight.

Routine 3: “I have bands!” 2 x per week to make progress, 1 x per week for maintenance.

One of the first pieces of equipment I bought was resistance bands. They’re super cheap – you can usually get a set for $20-$30. I bought them because I wasn’t sure that I was actually going to stick with it enough to make it worthwhile to buy weights, which tend to be much pricier. (Note: I don’t have a *lot* of weights, and what I have are the least expensive I could find, but I’ve still spent about $200-$250 on them over time.) So, they’re a good first step while you’re trying to build momentum. They’re also very portable, so good for traveling.

Routine is simple: follow routine 2’s structure, doing resistance band squats, chest presses, and rows, progress using the same rules as in Routine #2.

Various Notes

(1) These routines are designed to provide SOME results with a minimal time commitment. They are NOT optimized for results.

(2) For better results: (A) add in two more types of exercises – a vertical pull (pull up for example) and a vertical push (like a shoulder press), (B) increase the number of sets. There seems to be some evidence that 4-6 sets per muscle group per workout is good, with the goal of getting 10+ sets per week. While there is some debate about this, there is some evidence that “overtraining”, at least in terms of number of sets per week, is not a real thing, as improving results have been documented all the way up to 45 sets of an exercise each week. There are, however, diminishing returns for most exercises it seems. So, while 10 sets is better than 9, the benefit of bumping from 4 to 5 is greater than bumping from 9 to 10. There is also *some* evidence that there may even be negative returns in sets per WORKOUT past a certain point. But, that’s unclear. In brief: it seems fairly clear that the best way to add sets is to spread them out over time in any case.

(3) I know there were a couple comments related to this paper that the paper published. Unfortunately, it’s the weekend and I don’t have access to the journal from home, and those comments are locked. I’ll want to check those at some point.

Inflation and Wealth

~ 1250 words, ~7 min reading time

There’s been an interesting run of the media publishing articles about how great inflation is for ordinary people and how it hurts the rich. As an economist, these strike me as *VERY* weird articles. So, let’s flesh out the argument and what I think is wrong with it.

Basic argument: inflation is good for debtors, bad for creditors

This is correct, but leaves out some really important stuff. So, let’s consider 4 hypothetical people, and show how inflation affects them.

Person 1: Drowning in Credit Card Debt

This person is in very bad financial shape. Perhaps they lost their job during the pandemic, and could only find something paying much less, but didn’t manage to bring their expenses down. This person is renting, and has a credit card balance. They’re currently paying 12% interest on the credit card. They have, however, finally managed to achieve a “primary balance”, so, if it weren’t for credit card payments, they’d be just barely getting by. For these calculations, I’ll ignore that credit card interest compounds, so my estimates for credit card debt a year from now are a bit too low.

Before inflation:

Monthly income (after taxes): $2000
Monthly expenses: $2200
Credit Card Debt: $10,000 (this is growing by ~$300 per month, $100 from interest and $200 from additional charges)
Credit Card Debt 1 year from now: ~$13,600, about 6.8x monthly incom

Inflation happens: raising expenses (except credit card payment) by 6%, and wages by 4%. Interest rates on credit cards also rise to 15%. (Wages have been lagging behind inflation recently, and credit card interest rates have gone up.)

After inflation:

Monthly income (after taxes): $2080
Monthly expenses: $2320
Credit Card Debt: $10,000 (this is growing by ~$365 per month, thanks to a bigger gap between income and expenses and higher credit card interest charges)
Credit Card Debt 1 year from now: $14,380, about 6.9x monthly income

So, in comparison to income, the debt-laden person found their debts get *heavier*. This seems to contradict the basic argument. However, the basic argument assumes *fixed* interest rates (see: most mortgages and car loans in the US). With variable interest rates, interest rates will adjust to at least partially reflect the increased inflation rates.

Person 2: Living nearly paycheck-to-paycheck.

Here we have a person who is basically living paycheck to paycheck, but manages to save about 5% of their income in an emergency fund. Their income is the same as the previous person, but they don’t have the debt burden, and have a better balance between income and expenses – this may reflect that Person 1 was probably unexpectedly earning less than they were used to, while Person 2 has had time to adjust their life to their income. Like Person 1, this person is a renter – so all of their expenses are subject to inflation.

Before inflation:

Monthly income (after taxes): $2000
Monthly expenses: $1900
Accumulated savings: $1000 in a savings account earning 0 interest, saving $100 per month.
Current Savings = 53% of monthly expenses
(This was pretty close to “me in grad school before I got married” – though I did have some investments.)

Then, inflation strikes. Prices go up 6%, but wages go up 4%. (Recently, wages have been on the way up – but have been lagging behind inflation.)

After inflation:

Monthly income (after taxes): $2080
Monthly expenses: $2014
Accumulated savings: $1000 in a savings account earning 0 interest, saving $66 per month.
Current savings = 49.7% of monthly expenses

So, this person is *clearly* worse off. Their ability to save has been cut by 34%, and the value of their savings has been eaten into, making that emergency fund less of a cushion than it was before.

Person 3: Middle-class with Mortgage

Now, we have someone who earns a bit more, and is setting aside significant savings (20% of their income). But, like most people in the middle class, their main asset is their house. Their income comes from wages or salaries.

Before inflation:

Monthly income (after taxes): $5000
Monthly house payment: $1000
Monthly other expenses: $3000
Monthly expenses (total): $4000
Saving: $1000/month (20% of income)

House: $150,000
Savings Acct: $20,000
Stocks: $100,000
Total: $270,000 – 67.5 months of expenses

(Note: I’m ignoring the mortgage itself here, since its value is fixed. Also, if that house value looks low – remember, I live in NE Ohio. Houses aren’t nearly as expensive here as most places.)

Inflation happens – 6% increase in “other expenses”, 4% in pay. House prices follow inflation rates, and stocks do better than that, earning 20% (approximately correct for this year).

After inflation:

Monthly income (after taxes): $5200
Monthly house payment: $1000
Monthly other expenses: $3180
Monthly expenses (total): $4180
Saving: $1020/month (19.6% of income)

House: $159,000
Savings Acct: $20,000
Stocks: $120,000
Total: $299,000 – 71.5 months of expenses

Here, the result is a bit less clear – in terms of assets, the person is clearly better off, judging by months of expenses covered by the assets – because stocks did so well. However, saving has become slightly more difficult. But, thanks to the mortgage payment being fixed, they can at least save more in dollar terms than before, unlike Person 2 who didn’t have the benefit of any of their expenses being fixed.

Person 4: An Independently Wealthy Person

Here, I’m imagining an independently wealthy person. They have a bunch of assets that they can live off of – so they don’t have any “earned” income. Let’s say that it breaks down roughly like the overall holding of assets for US households as a whole. (This might seem like a weird assumption – but we have quite a bit of wealth inequality, so it’s not *way* off to do things this way.) 30% is in nonfinancial assets that will follow the inflation rate (6%). 15% in cash and bonds with fixed values, and 55% in stocks and business equity that outperforms inflation (20%). I’m going to arbitrarily say that they have assets = 100 months of expenses to start out, and that all expenses are affected by inflation.

Before inflation:

Monthly expenses: $100,000

Nonfinancial Assets: $3,000,000
Cash/Bonds: $1,500,000
Stocks/Equity: $5,500,000
Total: $10,000,000 – 100 months of expenses

After inflation:

Monthly expenses: $106,000

Nonfinancial Assets: $3,180,000
Cash/Bonds: $1,500,000
Stocks/Equity: $6,600,000
Total: $11,280,000 – 106 months of expenses

So, this person is certainly made better off, though, measuring *just* from how much assets/expenses increased in % terms, the middle-class person with a mortgage benefited slightly more (about a 7% increase v. a 6% increase), thanks to the fixed part of their expenses.


While I developed all of these examples in good faith, we shouldn’t take them *overly* seriously. For example, while I do think that, in general, the wealthy gain from inflation while the poor lose, these illustrations were intended instead to make a few points:

(1) People with fixed expenses gain from inflation. In our example, that was just the middle-class with a mortgage. Everyone else was made worse off, or, in the case of the wealthy, had their gains offset, because of inflation hitting expenses harder than wages. (Aside: Robert Kiyosaki has suggested that really poor people don’t have debt, since they can’t qualify for it. I think this isn’t actually true now – instead, the truth is that poor people have *particularly bad*, variable payment debt.) It’s not as simple as “debtors gain”, since some debts have variable interest rates which tend to rise in inflationary environments. So, the *type* of debt matters.

(2) People with assets with variable values can gain from inflation. In this case, that was “stocks” – which benefited the wealthy and middle class particularly.

It is true that, for any particular fixed payment debt, the person who is paying it is made better off if inflation happens, and the person receiving the payment is made worse off. But, we shouldn’t be too quick to assume that the debtor is poor or the creditor rich.

Another Workout Update

~ 750 words, ~4 min reading time

Some experiments just don’t work, even if they seem to make sense at the time. That was the case for my last workout update.

Turns out there were two fatal flaws in this routine:

(1) The circuit routine made me run out of breath – not like I was gasping, but breathing was placing limits on performance that were totally unrelated to strength limits.

(2) Progression was basically impossible. By grouping exercises, it meant that I could only progress on, for example, floor presses – if I ALSO progressed on rows. Of course, these two are pretty unrelated. There’s no reason for the two to progress at the same rate. Also, there was a regression bias built in – if I failed, I decreased weight. But, this applied to ALL of the exercises with the same weight.

These two ended up slaughtering my motivation, so, even though the routine was quite short, I didn’t do it consistently, and was looking for excuses not to do it.

Some current thinking:

(1) Full Body Routine is out. Doing them the way I had for quite a while takes too long. Doing it the circuit way was also not workable. Consistency was dead for both.

(2) I like the idea of a 2 day split. I hate the idea of an Upper/Lower split for a simple reason. I hate legs. I know they’re important, so I’ll include them. But, *only* legs for a day is just a way to guarantee I stop. So, “Push”-“Pull” as the basic breakdown. Each composed of 3 main exercises. Push: dumbbell squat (legs), floor press (horizontal push), overhead shoulder press (vertical push). Pull: pull ups (vertical pull), dumbbell deadlift (legs), rows (horizontal pull).

(3) After some reading, it seems that 6-8 sets per workout for a muscle group is good for most exercises. This required adding some additional isolation exercises. (Though I didn’t do this for legs for purposes of motivation.) So, I do 5-6 exercises in each routine. (Generally, “more is more” is true, but there seems to be diminishing returns that set in pretty quickly, with the potential for negative returns.)

(4) There’s some evidence that going very high volume may help if you have a muscle group that is behind. At the moment, that’s my upper arms (also forearms, but I’m focusing on upper arms first). So, I’m planning to workout both biceps and triceps every day, either as part of compound movements or isolation, for a total of up to 30-40 sets a week. So, on those areas that need particular work, use “more is more” to your advantage.

(5) Long rest times (2 min) are beneficial. So, I’m implementing those for the core compound exercises.

(6) However, there’s good evidence that drop sets (without rest!) are nearly as good as regular sets. So, I’m using drop sets for isolation work to help save on time. Basically, a drop set is one where you drop 20% of the weight and immediately do another set as much as you can. Then, drop another 20% and do it again.

(7) Each exercise will progress individually and based on performance.

(8) There’s pretty good evidence that the number of reps don’t matter that much. So, I’m going with a fairly high number of reps (10-12) so that I can skip warmups, saving time.


Push: Squats 3×10, Floor Press 4×10, Overhead Press 4×10, Flyes 4×12 (3 drops), Curls 4×12 (3 drops), Lateral Raises 4×12 (3 drops) Muscle group breakdown: upper legs – 3 sets, chest – 8 sets, shoulders – 8 sets, triceps – 8 sets, biceps – 4 sets

Pull: Pullups 5 sets – varied reps (I have a pull up program I’m following), Romanian Deadlift 3×10, Rows 3×10, Calf Raises 6×12 (5 drops), Tricep Extensions 4×12 (3 drops) Muscle group breakdown: upper legs – 3 sets, calves – 6 sets, back – 8 sets, biceps – 8 sets, triceps – 4 sets

Progression Rules:
For core exercises, do as many reps as possible on the final set, maxing at 20. For drop set exercises, AMRAP the first set. If I beat the prescribed reps by 2, increase weight 1 step (step depends on exercise). If I beat it by 5, increase weight by 2 steps.

Deload rules:
For core: if I fail on set 1 or 2, drop weight 2 steps. If I fail on set 3+, drop weight 1 step. For isolation: if I fail on set 1, drop weight 1 step.

I tried both of these routines over the past couple of days. They each take about 45 minutes. Not too terrible. Here’s hoping that I manage to adhere, and see actual progress!