By Steven Englander, head of research and strategy at Rafiki Capital Management
I propose a microeconomic rationale for why macro wage performance is so weak, despite tight labor markets. The idea is that we are getting paid less for our job-specific knowledge because technology is making it easier to replace us without major loss of productivity with less skilled workers. The implications for markets:
- Flattish Phillips curve and low wage inflation continue for an indefinite period
- Living standards may increase because of lower price relative to wages, not higher wages relative to prices
- Monetary policy will have to get on with dealing with a low inflation economy -- this means setting aside obsessions about balance sheet reduction and setting up the facility to use fiscal policy as needed when the zero bound is approached
- It’s relatively positive for equities in innovating sectors
- Long-term bond yields will be driven by monetary policy fears, not long-term inflation worries
- Short-term policy rate moved will be capped by the sensitivity of the economy to interest rates which may not be large. Note that this cuts both ways – both tightening and easing may be ineffective.
The thought experiment
My idea is that wages are driven by how scared your boss is that you are going to leave. If replacing you, retraining your successor and waiting for him to climb the experience curve is costly, he will pay a lot to keep you from leaving. If you are a cog in a wheel, then he won’t care much.
Imagine an economy of a bus driver, a taxi driver, a cook, a translator, a baby sitter, a doctor and a foreign exchange strategist.... Conceptually you can measure average job specific content by asking the following question: if you randomly reallocated jobs among these workers how much would productivity fall? For example, if the FX strategist was given the cook’s job and the cook became a doctor and the doctor a taxi driver and so on, what would happen?
Say in Economy A, there is specific knowledge or character traits needed: a bus driver needs the specifics of driving a bus safely, a taxi driver knowledge of the street grid, the translator an excellent command of relevant languages, the baby sitter some proven degree of responsibility, the cook of recipes and technique, the doctor the body of medical knowledge, the FX strategist how to say ‘current account’ and so on. Now imagine the chaos and productivity loss, if the random reallocation occurred and none of the occupants of new jobs had the required skills.
Now, say in Economy B, the bus is programmed to avoid dangerous manoeuvres, all taxi drivers have a GPS (unlike NYC where none seem to), the translator has automated translation at his fingertips, the baby sitter is aware the house and liquor cabinets are cameraed, the cook has a set of packets to mix (or almost equivalently the packets are sent to your home for you to mix), the doctor a diagnostic program and the FX strategist a chatty virtual assistant that can say ‘current account’. If a random job reallocation occurred in this economy the productivity loss would be much less. My conjecture is that wages would be lower
because there would be no need to bid to retain workers if they were readily substitutable, or if the same jobs could be filled with less specialized workers with no major productivity loss.
Wage compression is very likely to be a feature in Economy B relative to Economy A – that is, the premium one receives for job specific knowledge and experience would fall. If you throw in a bit of capital saving technological progress from the sharing economy and economies of scale from the low marginal cost of replicating many IT-based innovations, you could end up with a kind of immiserization of parts of the skilled and semi-skilled working classes.
Evidence is partial, but it is not straight forward to test this speculation. Figure 1 shows wage levels in selected industry groupings. Note that wages in motor vehicles and parts (bright blue) started way above over industries, but is now average for durable goods (red line) and below education and health services (green line) which started way below. Motor vehicles and parts are now way below the average wage in the private sector, having started above 50% higher in the 1990s.
In Figure 2 we index these industries to 100 in 2000. We note that wages in leisure and hospitality (light blue) and education and health (green) have both grown faster than in durables manufacturing (red) and above the average for all private industries. Vert similar patterns emerge if we index to 2011.
So wages have grown slower in high paying industries, faster in low paying industries and the net is the mediocre observed wage growth. What isn’t consistent is Atlanta Fed wage evidence that suggests the quit rate is back to normal for this time of the cycle and the wage premium for quitters is as high as it was in the early 2000s. The other side is that the Atlanta Fed data shows the gap between increases of skilled and unskilled workers as having narrowed.
The problem for central banks is that we know little of what triggers such shifts in labor market power, how long they last and what ends them. As long as these shifts persist, the Phillips Curve will look flatter in two-dimensional Unemployment Rate/Wage Inflation space. A well-specified wage equation that account for such structural changes would have a steeper inflation/unemployment trade off than one without the term but capturing the effects we discuss above is not so easy.
The type of technological progress would imply lower price pressures because the wage weakness would be transmitted in part into prices. (Full disclosure, you have to believe that there is an unmeasured component of actual productivity change here, although it may show up as quality-adjusted labor productivity, rather than standard output per-worker or worker-hour).
These disinflationary pressures may be hard to fight. Combine this with investment that is not overly responsive to interest rates and you have a situation where getting the inflation that you want may be impossible without risking undesirable levels of asset price inflation. It sounds as if the Fed is already there. There is nothing inevitable about this outcome, but it emerges easily if the disinflationary pressures are strong enough and the interest rate responsiveness is low enough.
One policy response is to live with it. Ultra-low inflation countries such as Japan and Switzerland have done just fine by many measures and the zero bound becomes an issue in a recession, not during an extended recovery. By ignoring it you have some ability to rein in asset market exuberance, but you are compromising on inflation and possibly activity targets.
This does not necessarily stop you from raising rates, but you are faced with a dilemma. If raising rates is effective you end up with the downturn you wanted to avoid, if raising rates is ineffective you are fooling
yourself in thinking that the margin versus the zero bound means that you are all clear in the next downturn. Being able to raise policy rates to three percent without tanking the economy very likely means that you can cut them in the next recession, you won’t have much of an impact. Hawks would argue that reducing the risk of a financial market bubble reduces the risk of a recession down the road.
It seems to me that whichever way you turn, fiscal policy has to be taken out of the doghouse. Post election, Fed officials reversed their pre-election love affair with fiscal policy, arguing that the economy does not need it. One might say that if the inflation undershoot turns out to be persistent, fiscal policy will become even more necessary to offset structural pressures. And if you think that high liquidity is contributing to asset market ebullience, then a bit of fiscal stimulus combined with monetary tightening can maintain activity and unwind some of the asset market pressures.
Modest long-term price pressures are probably positive on the fixed income side. Long-term disinflationary pressures and modest investment will keep downward pressure on long-term bond yields even if the Fed tightens at the short end in response to fiscal policy. The use of fiscal likely means being more relaxed about the size of the balance sheet. Debt-to-GDP would have to grow both cyclically and structurally, but debt servicing may not grow very rapidly because of the low inflation and the Fed’s interest income being recycled back to the Treasury. Short-term policy rates may swing around a lot versus stable but relatively low long-term rates.
The central bank could take a hard line and maintain or shrink the balance sheet even as fiscal expansion was put in place. Still, it would hardly help macroeconomic stabilization if government finances were called into question, so willy-nilly it is likely that the balance sheet would absorb some of the debt incurred via fiscal policy.
Caveat emptor – this analysis is pretty long term. In the short to medium term, I expect central banks like the Fed and ECB to try and follow up their rhetoric with liquidity tightening and for this to be reflected in long-term rates. Only if/when it turns out that the tightening is unsustainable will the forces I discuss above come into play.
On the equity side, if you can replace a skilled worker with a less skilled worker that is an attractive proposition. It is not as exciting as booming demand but it still reaches the bottom line. The social consequences are mixed. It is possible that this reduces the returns to certain types of training and education, both specialized and generic, but overall demand for relatively undifferentiated blue-collar labor will go up as will their wages. Improvements in living standards are likely to come via lower prices than higher wages. This is hardly the American dream. However, it is often difficult to put together policies that efficiently offset technological forces to provide distributional equity, and if other jurisdictions are not so focussed on distribution, you can end up with the worst of both worlds. It is possible that workers will drift to occupations where differentiated skills can earn a higher return – so maybe fewer doctors and lawyers but more dancers with the stars.
If you look at any central bank econometric model, the demand side has decades of development, the supply-side and particularly the modelling of technological change is primitive, distribution is virtually nonexistent and asset market bubbles a problem because they should not exist in the model world. These secondary issues have become first order issues. Unaddressed they mean incomplete policy regimes and surprising and disappointing outcomes.