This is the sort of story that feeds very nicely into the narrative that production moves to places with low wages, but I think if we think about this carefully, we see that that narrative is incorrect. We need to distinguish between changes over time and comparisons at a point in time. We also need to distinguish between corporate returns on a given project and perpetual returns on reinvested capital.
Changing Geography of Production
I have made this point before. It is not true that production moves to places with low wages. This should be obvious in many ways. Production happens overwhelmingly in high wage economies. If we ordered nations from highest wages to lowest, most of the bottom of the list would be nations that are not attracting capital. In fact, expanding production is not related to low wages. The Congo, in its current form, could have low wages for the next 50 years, and we will not be importing semiconductors from them. The only future scenario where we would be importing semiconductors from them, would be a scenario with higher Congolese wages. Expanding production is related to rising wages. It only looks like production moves to countries with low wages because the surest way to have rapidly rising wages is to begin with very low wages. Low wages are readily noticeable, so that is what we notice, and we establish a false sense of causation.
The causation comes, more precisely, from improving institutions, which cause both the inflow of capital and the rise in wages. So, places such as Taiwan and South Korea are high wage nations today even though it seemed as though they had attracted capital by being low wage nations. As growing nations expand production, they naturally develop competitive advantages in a wider array of productive areas. This happens most strikingly in nations playing catch-up, so at any moment in time, production appears to be moving from high wage to low wage nations.
This misidentification comes also from the tendency of developing economies to expand into established productive areas while developed economies expand into productive frontiers. So, assembly lines move to China, but Apple and Google headquarters locate in the US. The transfer of existing production is more palpable to us than if Chinese expansion had simply grown on the frontier because of the visible dislocations that it causes. But the flows we see are probably inevitable since activities on the technological and productive frontier will come from the most developed and secure economic areas.
If a genie offered me wishes, one would be for economists, especially ones that should know better, to refer to these movements of capital in these terms - as production moving to places with rising wages, not as production moving to places with low wages. At a shallow level, this is counterintuitive, so using the incorrect colloquial terminology is easy, but the truth of the matter is obvious, and I think, after all, not controversial. And its proper identification is important.
Improving institutions (liberal capital and labor regulation, property rights, legal universality, etc.) lead to rising production and rising wages. And, countries with better institutions and higher per capita GDP have higher labor compensation as a share of GDP. This is because, first, the inflow of capital and the improvement in local institutions increase local productivity, which leads to higher GDP. But, a secondary effect is that more economic stability means that corporations require lower returns to achieve the same risk-adjusted rate of return, and it is the relative risk-adjusted rate of return that influences capital flows. Less capital risk means higher labor compensation as a share of GDP. I think we even see this over time in the US. (Here is a new paper from Robert Lawrence, making the case that higher compensation share comes from increased capital investment because capital & labor are complements. [HT:MR])
Changing Incomes During Business Cycles
Another source of misidentification here is that corporate owners overwhelmingly bear the burden of cyclical volatility. So, we identify changes in economic activity with corporate profits.
But, these are movements in and out of disequilibrium. The movement of production to developing economies is not a cyclical movement. There may be a component that looks cyclical, if there is a lag between capital inflows and rising wages. But, the enduring effect is to increase the rate of income gains for all of the income categories.
Perpetual Corporate Profits vs. Project Profits
I think there are two things happening when a firm moves production to a rising wage economy. (I first typed "low wage". Even I am not immune.) First, because of the higher risk to capital in an economy that has not fully developed trust, corporations will require a higher return on assets. This is the factor that leads to a lower labor compensation share in developing economies. Some of these risks are political, and some of them play out as excessive cyclical risks, such as what we see today in China. To an extent the low institutional trust feeds excess cyclical volatility, as Tyler Cowen touched on recently regarding China. So, there are some excess returns that reflect long-tail risks, and some that are conventional returns on excess volatility.
In developing markets that are functional, these risks amount to just a few percentage points of net returns, which I think could explain the roughly 10% difference in compensation share between developed and developing economies. (If a 7% real return on corporate assets amounts to a 35% income share in developed markets, the 45% share in developing markets would be associated with required real returns of around 9%.)
But, there is a subtle issue of appearances here, too. The lack of safety in developing markets shortens the expected life span of intangible assets. Thinking of the comment that began this post, the firm is making a tradeoff, which is really imbedded in this trust issue. So, we might think of this in terms of project-level cash flows. If a project includes a billion dollars of assets with a 20 year lifespan, and we think of a simple required payout model, then a 7% required return would need $94 million in annual cash flows. A 9% required return would need $109 million in annual cash flows. Not that different. But, what if the lifespan of the assets is reduced to 10 years because of the vicissitudes of operating in a developing economy. Now the required cash flows for a 9% return are $156 million. This makes a huge difference.
Higher depreciation appears to be a factor in low productivity economies. Is this a significant source of the high capital consumption that could be at the root of declining global labor shares? Could the global footprint of US corporations mean that some of this is seeping into inputs of US capital incomes? This seems likely. We might think of this era of globalization as having some of the characteristics of the fashion industry. Innovators must constantly re-invest to stay ahead of the copycats. Maybe much of the intangible value in Apple, Google, etc., is not that different from the intangible value of, say Gucci. The emergence of a not-yet-fully-developed global economy makes the present context inevitable.
But, for my purposes here, this has a significant effect on the appearance of profitability of operations in low wage economies. These shortened capital lifespans mean that more reinvestment is expected. So, higher depreciation means that the cash flows of a single project would be much higher, but the cash flows of a firm operating in perpetuity would still simply reflect a 9% return on capital. Yet, if we are only looking at the cash flows of a specific project, it would appear (in this scenario) that a firm that offshores their operations is increasing cash flows by 60% by moving to a low wage economy. This is an exaggeration of the profitability of continuing operations, though, even after factoring in the higher required return. As I noted above, the difference in capital incomes between the developing world and the developed world can be accounted for by a relatively small increase in required returns.
Depending on how these risks are accounted for (much of this may be related to off-balance-sheet intangibles), accounting profits may look high initially after a move. At the same time, relative wage levels in developing economies are even lower than the total compensation share, because of lower productivity among the workers. These lower individual wages are not reflected in lower total labor expenses, so they are unrelated to the firm's investment prospects. But, the simultaneous association of those low wages and high current cash flows for the firm to new developing economy production might also create an inflated sense of a profit/wage transfer.
Labor is the Residual Claimant
Given all of this, let's think about the factors involved in offshoring an operation. For simplicity, let's think about this in a zero growth, zero inflation context.
An operation in a developed economy, with 7% required returns:
For total gross annual production of $250 million.
$1 billion asset value with 20 year lifespan
$50 million annual reinvestment for capital consumption
$70 million annual net operating profit
$120 million annual cash flows
$130 million annual labor compensation (based on 65/35 labor/capital split)
$250 million annual gross production
Now, using this simple model, I can estimate the labor/capital split for a set amount of production, given a required return and a depreciation rate. Let's assume the extreme example above, where asset lifespans are only 10 years - half the developed economy lifespan. What if required returns were 10%.
$100 million annual reinvestment for capital consumption
$100 million annual net operating profit
$200 million annual cash flows
$50 million annual labor compensation (= 33/67 labor/capital split)
$250 million annual gross production
This is the capital split of some of the least developed economies. Here is a table of estimated labor shares from this extremely simple model. Conceptually, we should expect lower labor shares from both higher required returns and lower asset lifespans.
The relative level of wages per capita is much lower in developing markets than the relative total labor compensation level. That reflects labor productivity issues beyond the scope of my analysis here. Although, the fact that the very low individual wages are not a product of low national labor compensation share also points to the fact that low individual wages are not the result of a transfer to corporate profits. This point, itself, separate from my analysis above, belies the notion that firms gain profits from moving to low wage economies. In fact, considering the huge difference between incomes in a country like China and a country like the US, the lower productivity of individual workers in low wage countries must be a much larger explanation of their low wages than the low productivity of capital or the higher required return on that capital. As an explanation for low wages, my analysis here of the capital decision must be a relatively small part.
But, regarding the main point, it should be clear that the lifespan of assets within an economy and the expected revenues from an operation are fixed from the perspective of a firm making a location decision. And, the target, risk-adjusted required return is also exogenous. In any individual investment, a firm will aim for positive net present value (NPV), but over time unless the location of operations has erected barriers to entry, aggregate returns will be bid down to the required return level. (And, in the paradigm we are using here, limits to entry are themselves an example of poor institutions.) So, the secular residual claimant to revenues from production is labor.
If institutions of a given economy improve, this will be reflected either in lower required returns or better asset utilization. Capital will be attracted to that economy until wages rise to the new equilibrium. Wage levels are a direct result of these interactions. From the perspective of a firm, the only variable here that isn't fixed is the wage level.
As we saw in the second graph above, cyclically, equity is the residual claimant. In other words, when there is a cyclical shock, profits are what changes. They are the only class of income that can change cyclically. But, at the secular scale, when there is a permanent change in the perceived risk of investment, labor is the residual claimant. In other words, when there is a secular change in context, labor compensation is what changes. It is the only class of income that can change secularly.
* I should note that while higher wage nations tend to have higher compensation share of income, there has been an international decline in compensation share over the past 50 years or so (pdf, pages 41-44) in both high and low wage countries. This is generally blamed on the new labor pools in developing economies bidding down global wages. I think this explanation captures too much support because of the framing of global incomes I described in this post. Labor is the residual claimant over the long term. Wages are the effect, not the cause. So, I think this is more of a mystery than it is generally seen to be. The housing issue may solve some of the mystery of what is happening here. Since capital gains are generally not included in BEA income accounts, the marginal shift of a household from renting to owning creates complex shifts in reported income. Homeowners tend to have larger housing expenditures which are recorded as income to financial enterprises, because a nominal mortgage includes an inflation premium. The interest payments created by that inflation premium are counted as capital interest income, but the capital gains of the home are not counted as household income. The marginal shift of a household from renting to owning is usually associated with a larger cash expense, because much of the mortgage payment is really a saving vehicle which is manifest through rising home values over time, due mostly to inflation in the long run. An owner-occupier household will generally have higher cash expenses than a renting household, at least at the outset, when there is a significant mortgage. So, there is a large transfer, in the data, from rental income share to interest income share (because the inflation premium is subtracted from rental income), and also a small transfer from compensation share to interest or profit share (because a leveraged owner-occupier has higher cash outlays than a renter does, since much of the benefit to a homeowner comes from expected capital gains). Globally, as national economies develop, homeownership tends to rise, and nominal expenditures on housing tend to rise in general. These trends would tend to increase this bias in income reporting. Additionally, this inflation premium was very high in the late 1970s and 1980s, which is when much of the downward shift in labor share occurred (especially in developing economies). Homeownership rates in the US are actually relatively low, so this effect is global.
This is only a hunch - I haven't done the difficult work of digging deeply into global income data. If global trends match US trends, the increase in homeownership has also come with higher mortgage leverage levels, especially as populations in developing economies become more urban and move from basic rural homes to professionally built homes. In the US compensation shares graph shown above, the high Rental Income share (which is to owner-occupiers) before the mid-1960s was during a time when homes were less leveraged, so these distortions weren't as pronounced before the inflationary period. As I have argued in my housing series, imputed rental returns are currently very high relative to mortgage rates, so homes are very deleveraged now compared to their intrinsic values. This is why rental income is rising again. One can argue either that this is a product of a disequilibrium or that homeowners are earning a higher return (on a lower intrinsic value) because homes now have a higher perceived risk of volatility. Whichever framing one chooses, we can see visually in the graph of compensation shares above how secular changes in compensation share are related to relative returns to homeowners and landlords. Total corporate profit, interest, and proprietor returns have not moved out of their long term flat range.