High Wages encourage Innovation?

High wages increase the benefits of an innovation that increases labour productivity. The higher the wage, the bigger the benefits of saving an hour of labour to produce the same quantity of goods. But if that innovation itself requires labour to think up the new idea, test it, and implement it, then high wages increase the costs of innovation too. If benefits and costs increase by the same percentage, high wages will have zero effect on the number of innovations that pass the expected benefits > expected costs hurdle, and so zero effect on the number of innovations that would be implemented.

Think of a simple economy where land and labour produce food. At the aggregate level, what matters is whether innovation itself is more or less labour-intensive than the economy as a whole (or those sectors of the economy to which innovation might be applied). Think of an experimental trying out a new variety of wheat using the same labour/land ratio as the economy uses for existing varieties of wheat. The cost of the experiment is the opportunity cost of the existing wheat that could have been grown on the experimental plot of land using the experimental labour. The expected benefit of the experiment is the increased yield if the new variety is a success. The relative scarcity of labour to land, and the relative wage/rent ratio, has zero effect on the expected net benefits of doing the experiment. What does matter is the real rate of interest (the nominal rate minus the expected inflation rate on wheat). Because the costs of the experiment come this year, but the expected benefits of the experiment come in many future years. It is low interest rates, and not high wages, that should encourage innovation. Innovation is an investment.

Sure, innovators are looking at the expected benefits of innovation, and don't know the actual benefits. So it is always possible that labour-saving innovations will have a higher success rate than the innovators expected, and land-saving innovations will have a lower success rate than the innovators expected. If so, higher wages (and lower land rents) may cause innovators to start looking for gold in the exact place where the gold happens to be hiding. Just by luck. But it could easily have been the other way around.

Suppose that innovation, by its very nature, is a very labour-intensive activity. That sounds plausible to me. It takes a lot of labour, and very little land, to do the thinking, and measuring, and recording, and whatever, to find a new variety of wheat that works better than existing varieties. In that case, high wages would increase the costs of innovation by a greater percentage than they would increase the benefits of innovation. So for the marginal innovation, where the Expected Net Present Value was just barely positive, would become negative if wages increased. So fewer innovations would pass the E[NPV] > 0 hurdle if wages rose. If innovation itself is very labour-intensive, high wages discourage innovation.

You want to talk about skilled labour vs unskilled labour? Sure. If innovation requires skilled labour and saves unskilled labour, then a fall in the ratio of skilled to unskilled wages (a rise in the ratio of unskilled to skilled wages) will encourage innovation. But talk about that ratio; don't talk about just "wages".

And don't get me started on "capital"-saving innovation, because "capital" is the time-structure of production. Whenever you start a production process that has present costs and future benefits, you have capital. And the innovation itself is a capital project. And the benefits of innovation will probably outlast any production process it modifies. So a fall in the rate of interest will promote all innovation, including innovation that is purely "capital-saving". And if the real interest rate is negative, so it is profitable to store wheat even if some wheat does get lost in storage, then innovation that lengthens the production process (is the reverse of "capital-saving") is a good innovation. It means you can store wheat for longer before it rots.

I know little about economic history, and even less about innovation. But something about this "high wages encourage innovation" meme doesn't sound right to me.

[Update: Take the simplest case: an innovation costs C units of labour to implement, and benefits by saving B units of labour per year forever. No other costs or benefits. The wage is W, and interest rate is r. The Net Present Value of implementing is

NPV = WB/r - WC = W(B/r - C)

You don't need to know W to tell whether NPV > 0. All that matters is whether B/r > C.

Line up all the possible innovations, from the highest to the lowest NPV (NPV on the Y axis). The NPV curve slopes down, and crosses the X axis, because some projects have negative NPV. We implement if NPV > 0.

An increase in W does not shift the NPV curve up. It causes the NPV curve to swivel clockwise, around the point where it crosses the X axis. It has no effect on the percentage of projects that are profitable. A fall in r is what shifts the NPV curve up.]

Comments

[Ralph Musgrave]

Good point.

[Lord]

Benefits and costs increase in the same proportion? If that were the case, there likely would be no innovation, at all. Yes, it costs to develop a more productive variety of wheat and to identify where it can be suitably employed, but once done can be used everywhere suitable. The field being invested is only a minuscule portion of what a successful innovation would become. In the same way, the innovation of an advanced piece of machinery automating a factory will be high, but if successful, will be duplicated widely and used broadly. Innovations that can't deliver their return are failures and no one would invest without expecting much more than what they invest. Risk being much greater than interest rates, a much higher return is demanded.

[AB]

You're right, high wages don't necessarily encourage innovation, there are multiple other factors that do that, but they obviously shift the NPV of labour-saving capital investments to the point that an increasing % of known projects become financially viable.

[Nick Rowe]

Lord and AB: No.

Take the simplest case: an innovation costs C units of labour to implement, and benefits by saving B units of labour per year forever. The wage is W, and interest rate is r. The Net Present Value of implementing is

NPV = WB/r - WC = W(B/r - C)

You don't need to know W to tell whether NPV > 0. All that matters is whether B/r > C.

Line up all the possible innovations, from the highest to the lowest NPV (NPV on the Y axis). The NPV curve slopes down, and crosses the X axis, because some projects have negative NPV. We implement if NPV > 0.

An increase in W does not shift the NPV curve up. It causes the NPV curve to swivel clockwise, around the point where it crosses the X axis. It has no effect on the percentage of projects that are profitable. A fall in r is what shifts the NPV curve up.

[TK]

Well, but if we have a simple fixed cost in addition to the labour cost for the innovation, a swival clockwise implies that more marginal innovation investments have positive NPV

[louis]

I think when people make the link between wages and innovation, they have in mind the Industrial Revolution. Wages need to be high enough in order for it to make sense to substitute coal power for human labor at the margin. Once technologies are introduced, a path is created where innovation is directed to improving those technologies that have already been partially adopted. This may not be generally true in all places and all types of innovation, but it has a logic to it.

[Nick Rowe]

TK: Yes, if the innovation costs both labour and land, but saves only labour, that would be correct.

louis: Yep. I learned from Gabriel Mathy on Twitter that I am (implicitly) criticising economic historian Rob Allen's theory of the Industrial Revolution. If the supply of labour falls, and the wage/rent ratio rises, there will be a movement along the production function towards more land-intensive methods. (Coal=land). But it is not obvious to me why this would incentivise research into new methods any more than the opposite movement.

[Julius Probst]

Well, there is a large economic history literature on the Industrial Revolution and labor-saving technologies. Robert Allen is the most prominent advocate of this theory. According to him, it's not so much about high wages per se, but instead about the relative price of labor that matters, i.e. the ratio w/r. If wages are high relative to the price of capital, the more it makes sense to adopt labor-saving technologies.
Hicks wrote about directed technical change and more recently Acemoglu has also written about it (https://economics.mit.edu/files/4126).
One of the crucial assumptions though seems to be that the aggregate production function is CES instead of Cobb-Douglas.

Update: failed to read your last comment at first

[TK]

I guess I am confused about why you are so critical of the unqualified statement that higher wages increase investment in labour saving innovations then. All it takes is some small friction - perhaps management only considers investment projects with positive NPV above some threshold - and the statement makes perfect sense even in the extremely simplified model you present. Surely the default assumption practically should be that such frictions exist rather than that they don't. We don't have to worry about multiple labour markets or other resources for the statement to be true, we just need there to be labour savings and labour costs and friction.

Frankly, this post is interesting because it has made me think that the statement is more robust, less dependant on specifics of skilled and unskilled labour than I did originally.

[Nick Rowe]

TK: "All it takes is some small friction - perhaps management only considers investment projects with positive NPV above some threshold - and the statement makes perfect sense..."

OK, that would work to save the old theory. But if you changed it slightly, so that management only considers projects where the Benefit/Cost *ratio* is above one plus some threshold, we are back to my result.

[Longtooth]

Having spent an entire career in R&D for high tech (computer industry) product innovations (both patentable and proppretary not patentable versions) I note that there's this little thing called probability of R&D effort producing an innovation which meets the criteria B/r > C. That is the investment in A which must precede investment in C. So the condition must be B/r > C+A, where A = wage and overhead rate of R&D investment.

In technology development, of all varieties of technology including agricultural, WiA is an ongoing cost.. not a one time event related to B/r. The success rate of investment in A determines whether B/r > C+A so that we can make the equation more precise by B/r > C + sA where s is the success rate over time

And if I'm not mistaken, and I might be, A is considered a capitalized investment... not expensed at actual annual costs, but amortized over n years (say 3 or 5). That means capital (as A) can be written off with capital investment tax advantage, reducing the actual cost of A for profit/loss purposes after tax.

If A is capitalized then the equation becomes B/r > C + sAr.

I don't think this can be simplified to equate C to sAr since C was defined to the the cost to implement which can only occur after sAr is expended.

It was my 30 years of experience that C is always << sAr so that the decision to expend A (and it's magnitude) was based on an estimate of s (success rate over time). But s also depends on the magnitude of A (labor costs = number of people employed in R&D x total professional wage overhead), capital equipment, offices and lab space.

In reality the decision for A depends on the relation of Ar to B/r for any given s.

[Lord]

Since NPV is proportional to W it makes sense to introduce the innovation where wages are highest (England vs France). You might wonder why they didn't just use less expensive labor instead and I think nations and taxes were probably a large portion of this. A highly successful innovation will also lower r making more innovations profitable. Risk is generally much greater than r which is why investment is not very sensitive to interest rates. The limitation is not likely a positive net return, but the amount of capital willing to be risked. An investment schedule is not the best way to look at this as most of the action is making investments thought to be impossible and unprofitable into the realm of not only possible but highly profitable, moving them towards the origin. Until known, most things seem impossible.

[Nick Rowe]

Lord: "Since NPV is proportional to W it makes sense to introduce the innovation where wages are highest (England vs France)."

Smashes head on keyboard.

[Market Fiscalist]

Your example seems to cover the decision to spend money on R&D into innovation - not the decision whether to implement innovations that are already known about.

Take a community of apple farmers. Each farm takes 10 workers to produce 1000 apples and has to pay 80 apples to each of 10 workers in wages. A robot manufacturer could make a robot that lasts a year and is the functional equivalent of a worker. It costs 85 apples to produce the robot so it doesn't get made. A bunch of worker migrate to a different region and the apple farmers now have to pay 90 apples to each worker in wages. The robot is now worth producing. (assume that apple workers and robot-producing workers are different sets of workers).

Higher wages have encouraged innovation.

What is wrong with this logic ?

[Nick Rowe]

MF: Your robot costs 85 apples, and apples require both land and labour, so your robot requires both land and labour. But your robot saves labour, not land. So yes, in your example, an increase in the wage/rent ratio would make the NPV of robots more likely to be above zero.

Now change your example to that robots are produced with labour only. In that case, the wage is irrelevant, since an increase in the wage increases both the costs and the benefits of a robot, in the same proportion. Only the rate of interest matters.

[Market Fiscalist]

OK, so I suppose my example is covered by your 'You want to talk about skilled labour vs unskilled labour? Sure. If innovation requires skilled labour and saves unskilled labour, then a fall in the ratio of skilled to unskilled wages (a rise in the ratio of unskilled to skilled wages) will encourage innovation. But talk about that ratio; don't talk about just "wages".'

But it still seems it would be valid to say "a rise in wages relative to all other costs will increase innovation" as long as you assume that the costs of innovation are not made up entirely of wage costs.

[louis]

Nick: "If the supply of labour falls, and the wage/rent ratio rises, there will be a movement along the production function towards more land-intensive methods. (Coal=land). But it is not obvious to me why this would incentivise research into new methods any more than the opposite movement."
I think a lot of the disagreement​ here is a function of ill-defined terms or maybe the need for a better framework for modelling innovation. You say that a change in cost of labor relative to other resources (coal) will lead people to adopt methods that utilize labor less and coal more. That's fine in a static equilibrium, but where were those other methods? Steam power was not widely understood and the world didnt have good techniques to harness it. The point was the methods needed to be dreamed up in order to be implemented, the goals were more likely to command the attention of dreamers in societies where labor was scarce, and the innovations were more likely to be adopted and spread where relative prices favored labor saving. Moreover in a dynamic sense, wider dissemination of phase 1 technologies inspires tweaks to improve those technologies and shows other dreamers what is possible - path dependence.
Ps note i use the term "dreamers" because I romantically believe inventors are not just laborers, not even "skilled" laborers. There is something qualitatively different going on, the skill is on a different axis.

[Longtooth]

Nick,
Innovation doesn't just fall from mana. Somebody pays for the express purpose of innovating -- it can be a more labor efficient mfg'ing process, or a single tool that cuts labor by 50% or 90%... or more to the point, an effort applied to innovate these examples. The effort may or may not succeed or may not succeed as much as hoped ... it is therefore a capital investment risk with success rate probability s and interest r on capital applied to that effort (A).

It sounds like you're ignoring the innovation costs and assuming you can pick some innovation that already exists for sale from off the shelf. That is to say, for example, a robot innovator decides to mfg'er them for sale at some price P (which doesn't appear in your equation). Whatever that price though it includes sAr but then it has to include sales, admin, & profit as well.

I'm only pointing out that the cost of implementing an innovation can't ignore the cost of the innovation itself... yet you don't include it. Why not?

[Tel]

Do you want robots to take all the jobs? Show your support for technological disemployment of humans with this stylish shirt. Available in men's and women's styles using premium materials. All at a reasonable price of only 1 hour of unskilled human labor.

https://teespring.com/shop/robots-for-a-15-minimum-wage

Hurry! Last days to order. :-)

If innovation itself is very labour-intensive, high wages discourage innovation.

Clearly it's not labour intensive. From historical examples, most of the major scientific and technological advances came either from single people, or from small teams. We can see the same thing today with tech startup companies generally being most innovative when they are small. That said, finding the right choice of labour for the job is quite hard.

You want to talk about skilled labour vs unskilled labour? Sure. If innovation requires skilled labour and saves unskilled labour, then a fall in the ratio of skilled to unskilled wages (a rise in the ratio of unskilled to skilled wages) will encourage innovation. But talk about that ratio; don't talk about just "wages".

I agree completely, but who was it saying this was a simple matter of "wages" anyway?

It's meaningless to talk about either labour or wages as some homogeneous entity, obviously there's always going to be some people earning more than others. If you want to look at the top-level tech entrepreneurs the potential wages are huge... admittedly many don't make it; but all the people doing it think they will be the one who DOES make it. Thus, from an incentive point of view the high dropout rate isn't significant.

MF: Your robot costs 85 apples, and apples require both land and labour, so your robot requires both land and labour. But your robot saves labour, not land. So yes, in your example, an increase in the wage/rent ratio would make the NPV of robots more likely to be above zero.

At the moment we have fossil fuels, which from the robot's point of view look a lot like an effectively infinite expanse of "land" (just happens to be underground but provided it makes electricity the robot doesn't care about that). Some say there's a long term consequence for using those fossil fuels, maybe one day they might run out or something, but you know there's always someone predicting the end of the world, isn't there?

Even ignoring the fossil fuels, in certain domains computers are significantly more efficient than humans once energy consumption is taken into account (let's presume energy is roughly equivalent to land). For example, we probably remember "Deep Blue" beating the grand master at chess, but these days CPU design and chess software has come so far that it's possible to beat a grand master with something running on an iPhone consuming a small fraction of the energy that a human would use for the same task.

[Nick Rowe]

MF: "But it still seems it would be valid to say "a rise in wages relative to all other costs will increase innovation" as long as you assume that the costs of innovation are not made up entirely of wage costs."

No. A rise in the wage/rent ratio will raise innovation if the costs of the innovation have a smaller wage/rent ratio than the benefits (savings) from the innovation. And reduce innovation if larger.

Longtooth: "It sounds like you're ignoring the innovation costs...."

What??? This whole post is about those innovation costs! And about whether those innovation costs have a higher or lower labour/land ratio than the benefits.

Smashes head on keyboard again.

[Longtooth]

Nick, you defined innovation costs as implementing innovations... there's a huge difference between the costs of implementing them and creating the innovations in the first place. Implementation costs are minor compared to the costs of innovating. I spent 30 years doing both. Get real.

Smashes head on table twice!

[Market Fiscalist]

'No. A rise in the wage/rent ratio will raise innovation if the costs of the innovation have a smaller wage/rent ratio than the benefits (savings) from the innovation. And reduce innovation if larger'

Got it: If a rise in wages causes the costs of innovation to rise less than the costs of other things then innovation will increase. Interesting side question: Won't there be a second order effect where a rise in wages will lead to innovation in the innovation business to make it less labor-intensive, which will lead to more innovation in the wider economy ?

[Market Fiscalist]

While I agree with everything you say in the post and in the comments , it still seems so empirically likely that a rise in the cost of unskilled labor will affect the costs of the innovation less than the costs of other things that the statement "a rise in the costs of unskilled labor will lead to more innovation" will be true almost all the time.

[Roger Sparks]

Interesting subject and comment. But let's not discount increased wages as a motivating factor for innovation.

Nick offers the simplified equation "NPV = WB/r - WC = W(B/r - C)".

The WB/r term is the only positive term. It only needs to be larger than C to make NPV positive. Notice that when r approaches zero, WB/r approaches infinity. It is not hard to make NPV positive under zero r conditions!

Also notice that wages close to zero would result in near zero NPV. By the opposite token, increased wages INCREASE NPV. Clearly, wages have an effect on NPV.

Where I DO agree with Nick, is when we say that RELATIVE wages do not effect the decision of embracing innovation. It is the inflation of wages that drives the r term into negative territory that drives positive NPV.

[Neil]

A rise in the expected rate of wage increase can drive innovation, right? Basically if the expected rate of increase is higher than the discount rate your firm uses, then paying wages now to save wages later is a good investment. (And a rise in current wages may increase expectations for the rate of future increases?)

[Nick Rowe]

Neil: yes, for a given interest rate. Another way to think of this is that the "real" interest rate that matters is the nominal interest rate minus expected *wage* inflation (as opposed to *price* inflation, which is how we normally define "real interest rate").

[Odie]

Sorry for this likely very naive question but what would the real interest rate be in the absence of innovation?

[Ralph Musgrave]

No one has noticed the elephant in the room: how exactly do you raise wages prior to trying to improve innovation? Can be done by ceasing to replace capital equipment as it comes up for renewal, but that’s not a good idea. Could be done by cutting down on amount spent on education: also not a good idea.

Could be done by letting the external balance go to pot and running up debts to other countries: also not a good idea.

To illustrate the impossibility of magically raising wages, take a simple subsistence farmer economy where the only product is corn. Every farmer (and adult members of each farmer's family) produce 1kg of corn per day. Now how do you increase that without first innovating? Can’t be done!

[Nick Rowe]

Odie: the real interest rate would probably be lower than it otherwise would be (less desired investment, and more desired saving, at any given interest rate). But it could still be positive, if people are impatient. It depends.

[Jeffrey]

What happens to the theory if a significant fraction of innovation happens during leisure time?

[Nick Rowe]

Jeffry: it doesn't affect what I'm saying here. Either you enjoy thinking up new ideas and will do that anyway, regardless of costs and benefits, so Wage doesn't matter, or else you don't enjoy doing it, it's unpaid work, and the wage is the opportunity cost of sitting home thinking instead of going out working.

[Oliver]

If innovation itself is very labour-intensive, high wages discourage innovation.
(...)
But something about this "high wages encourage innovation" meme doesn't sound right to me.
Commenting only from local experience, not economic understanding: Switzerland has been suffering from the Dutch disease for more than 1/4 of a century now and my guess is that the type of innovation the high cost of labour (+ high cost of everything else) has encouraged, if any, is the type that cuts down on labour intensity (for any specific market). E.g., what used to be a large textile industry now consist mostly of niche producers who make high-tech materials (e.g. sails for Alinghi) or high end designer fabrics.

[rsj]

All this discussion seems a bit like medieval theological debates. We don't know what leads to "innovation" in a general sense, and the word itself is so abused, especially by economists discussing development problems. I don't see how any of these arguments can be taken seriously.

If this was a serious discussion, you'd look at history and see what happened to previous episodes of innovation. For example, the flourishing of printing in Venice, or silicon valley in the 90s and 2000s, or some other period associated to "innovation". None of these areas were low wage areas. But that assumes a serious attempt at finding out the truth, rather than a discussion about angels dancing on the head of pins, or some equally theological tale about expected net present value of returns to "innovating".

Just my opinionated view of silicon valley is that you need an eco-system: networks of venture capitalists, experienced business founders, a pool of newly minted college graduates, and a pool of businesses all ready to hire and acquire. You need these network effects in order to mitigate risks. For example, a venture capitalist isn't going to put all his money into one company, they need many companies, but all in the same social networks/geographic circles. Similarly, an employee is going to want to move to a place with many start ups, not just one, because when the one dies, they want to be able to jump to another. For the same reason, a company wants to locate to a place where there are already many experienced workers.

Once you have this eco-system, you need a market. The biggest problem is always not enough people with extra money to spend. We used to sail the seven seas in search of gold and silver, but now we search the world for customers. Maybe in Venezuela they are supply constrained, but everyone else is dying to find a customer. Specifically, since the marginal cost is almost zero, all the costs are fixed costs, and if your market is twice as large, you can spend twice as much on research and development. In the reverse, when people cut back their spending, it affects tech disproportionately because producing fewer copies doesn't save them any money. They have to cut R & D, and often massively, in response to a small decrease in demand.

So in that narrow view, raising average wages for everyone absolutely would help drive innovation in tech. Tech already pays very high wages and yet has huge operating profits. I suspect there is a similar dynamic at work in other types of publishing, for example historical publishing.

Whether that same definition of "innovation" would also apply to other areas with razor thin operating margins, such as farming, instead of margins of 60% as in tech, is another story. But my guess is that those industries where we see innovation are closer to tech than to farming, but I think Nick's intuition comes more from farming.

[Lord]

High profits/wages primarily encourage innovation by increasing risk capital, lowering the effective r. The key here is not high wages at the expense of profits, but due to them, so you need to know why they are high, and high profits are most important. High profits will self fund innovation. In contrast, innovation that requires high capital investment constrain other investment and are constrained by it. It still takes initial innovation to get the ball rolling, and the greater the innovation, the more profitable, the more innovation can be funded.

[rsj]

^^^ Lord, we are not talking about the wages of the tech workers, who are an insignificant fraction of the labor force, but of *all* wages. At any point in time, only a small slice of the workforce is engaged in anything that can be generously described as "innovation". At most, say 2%. When you raise all wages, you are primarily raising the wages of the customers of the 2%. Sure the wages of the 2% may also go up, but the revenue of selling to the 98% goes up much more than the costs of paying the 2%.

[Lord]

While higher profits don't always result in higher wages, higher wages are usually a sign of higher profits and higher profits lower the risk encountered and premium demanded. English wages and assumably profits were higher. The scope and size of the innovations were large, cloth particularly, which affected everyone, everywhere, and lowered costs exceptionally, only a portion of which was captured by producers. Initial wage growth was in population and scale rather than wage rates. Extensive innovation through scale is much easier than intensive through technique. I am not sure any recent innovations can compare but will contribute to more, both diverse and numerous.

[Nick Edmonds]

I've wondered about this too.

I think it's useful to think about causation here. The real wage (which is what matters here) is not itself endogenous; it depends on how firms set prices over labour costs, even if that's some form of MCP. We can't just assume a change in the real wage, all other things being equal. We need to know what is causing the change in the mark-up.

If, for example, something changes to increase the required return on capital - say an enhanced premium for equity risk - then we might expect to see both an increased mark-up leading to lower real wages and a reduction in innovation. This is possibly what we are seeing, but it would not really be a useful description to say that the change in wage level is causing the change in innovation.