EU growth: what happened?

The economic structure seems significantly skewed towards exports – in the current context of tariffs and retrenchment, the EU’s growth model is jeopardised. 

Euro Area growth is waning

Long a preferred measure of economic activity and inflation drivers, a plot of Producer Price Indexes shows significantly different paces of economic activity whether considering investment or consumption goods.

The EU economy is less competitive in consumer goods than in investment goods. More worryingly, the gap between domestic and total capital investment prices in the Euro Area is far more significant as a result of the explored comparative advantage. 

The Balassa-Samuelson effect could justify the gap, as could the Podpiera effect. Arguably, the widening gap could hamper investment, depressing endogenous growth and exposing the economy to foreign demand shocks.

Considering how the significant gap between domestic and total producer prices of investment goods disproportionately affects domestic firms, it is arguable that the EU’s industrial fabric is skewed towards serving (mid-tech) foreign technological demand rather than domestic demand for more advanced intermediate consumption (in trend since 1995). In that, less diverse supply hampers competition and produces the visible price gap.

Sales of High Technology components from/to High Income (HI countries) countries fell between 1995 and 2008, while it increased from/to Developing and Emerging Economies (DEE). Exports of high-tech products from EU-28 fell considerably after 2000 while the current account surplus increased significantly. 

The gap between prices for investment goods practiced domestically and abroad is far wider than say, in Japan.

This is worrying and structural and it significantly impacts firms’ incentive to invest and innovate. It’s not a single cause nor a cyclical cause – but it is there.

More to follow.

How to keep the EU economy growing next year

By stimulating capital investment. It implies adjusting capital depreciation rates and the deductibility of interest payments, both Union-level instruments. Such policies may deliver a more effective stimulus to investment than the ECB’s interest rate policy.
Amongst the EU’s most significant challenges in the coming years is arresting the sluggish/declining productivity growth of the past few years and return to income growth at (least at) the long-term pace of inflation target of 2%. 

As a job-intensive recovery depressed investment, despite the efforts of ECB and Commission, real disposable income grew at 55% slower from 2009 than in the nine years preceding that year.

labour market squeeze could threaten these gains in the years going forward. Productivity actually grew faster from 2009 than in the years preceding it, regardless of whether we measure it by employee or work intensity.

Table above: Labour Productivity indexed to 2010

The link between productivity and income could hint at higher capital intensity fostering productivity growth – or lower labour costs relative to the production horizon of capital and technology (in the neoclassical framework of Romer). 

Because investment measured by Gross Fixed Capital Formation as a percentage of GDP decreased 7% from 2000, productivity gains are most likely the product of a labour market rout (coupled with wage restrain) than of efficiency gains in how inputs are transformed. This hypothesis is coherent with slowing growth of real income per capita. 

Table above: Gross Fixed Capital Formation as a percentage of GDP

Imported inflation, the labour market squeeze or a demand shock could very easily erode the gains in productivity and income – if investment and innovation don’t resume.

The current series of posts questions the assumption that firms increase investment when interest rates are lower. Survey data and microeconomic research question this assumption too.

For instance, Acemoglu’s capital-augmenting technology function hints at lower incentives to capital investment from lower interest rates. 

Caballero defines a capital investment equation that considers labour-capital elasticity (or the incentive for firms to invest instead of hiring) based on capital cost (the so-called user cost) and capital intensity (or the amount of capital involved in production). Caballero argues that firms consider the current level of capital and the interest rate when deciding whether to invest, subject to other costs of investing such as availability of qualified persons to operate that investment. 

In practice, holding the value added of buying a new machine constant, a ten-fold decrease in the interest rate only increases investment by 25% controlling for adjustment costs. 

That is, higher adjustment costs affect capital investment. Typically governments let higher capital intensity increase adjustment costs as there’s only so much capital intensity a firm may accumulate before becoming costly to integrate yet more capital.

Governments may otherwise choose to affect capital levels through higher depreciation rates that increase the incentive to invest; or introduce incentives to investment that decrease the costs of capital such as increased depreciation rates.

These instruments may reveal essential when market interest rates increase from 2019 onwards, and may be activated at European Union-level through the CCTB, the Common Corporate Tax Base. The Union may adopt flexible calculations of the depreciation base of a leased asset (increasing or decreasing the incentive to investment, Article 32) and of the deductibility of interest (increasing or decreasing the disincentive to investment, Article 69, CCCTB) in the scope of these directives. Combined, both affect a firm’s decision to invest.

The work of Cummins and Hassett is undoubtedly valuable to understand the firm’s investment decision. Beyond finding a relationship between the user cost of capital and investment by focusing on tax reform years, the authors define the tax-adjusted user cost as increasing with lower interest rates (hence, the low interest rate environment hampers investment – as long argued in this blog – albeit lightly as the ECB agrees) and decreasing with lower depreciation rates.

In other words, lower interest rates should be combined with extended depreciation rates such that firms are enticed to accumulate capital. Otherwise, a policy-bound accelerated depreciation schedule accelerates firm investment decisions.

As such, in the years after 2019, the Union should focus on an accelerated, policy-bound accelerated schedule of depreciation to induce capital renewal followed by gradually increasing depreciation rates (to a steady state) that permanently decreases the capital user cost.

In practice, this implies the Union can consider the labour-intensive recovery when granting leeway to investment while preventing overinvestment.

Big data and self-driving vehicles

These two technologies represent the diminishing returns of investing in capital-augmenting technologies while labour-augmenting technologies remain competitively more productive – in different ways.

Big data is a capital-augmenting technology that does not compete but is actually complementary to labour intensity. It may be defined as a pivot technology between capital-labour substitutability and complementarity: whereas the technology’s development relies on extended human input, its applications shall undoubtedly replace labour-intensive tasks such as economic forecasters or warehouse clerks. It is a technology, not innovation.

Concurrently, self-driving vehicles are currently put off by the abundance of labour-augmenting technologies in the form of platforms that distribute work. This is plausible as capital and labour in the US remain complementary. 

Going forward, the trend for more intense work diminishes returns. To a firm in a competitive labour market, labour presents diminishing returns as higher human capital requires higher wages constrained by regulatory requirements to effort intensity. That is, a more able engineer may be unable to work longer while earning more.
Labour only displays increasing returns when complemented with capital (defined by Romer, freshwater economist, as knowledge – without distinguishing between labour- and capital-augmenting technologies), itself exhibiting increasing returns.

Labour’s diminishing (perhaps constant) returns implies the current job-based recovery has resulted in increasing nominal and real wages as the labour share measured in nominal terms increased.

A diminishing labour share in real terms (with such a gap between light blue and green plots) is explained differently by Keynesian saltwater and Sargent freshwater economists. 

Saltwater economists would perhaps argue that public stimulus increased demand and productivity, and tightened the labour market leading to higher inflation. Freshwater economists consider economic agents anticipated inflation and thus reacted to public stimulus by increasing wages and prices without a real effect on productivity. 

Plotting productivity 

the increasing gap between nominal wages and the purple plot, labour productivity, does not falsify freshwater’s perspective nor the observation that a (diminishing returns) labour intensive recovery does not increase productivity.

Structurally, it suggests that a recovery reliant on labour strength is inflationary not via demand but via supply-side effects as Janet Yellen so accurately explains: firms compensate higher labour costs resulting from lower productivity by increasing prices.

Going forward, the narrowing gap between the value of capital- (purple) and labour-augmenting (blue) technologies (plotted above) in real terms implies firms shall find comparatively less profitable to invest in capital – as long as interest rates remain subdued. Notice the gap was narrowest in the year preceding interest rate increases. Yet interest rates are bound to increase as the diminishing returns of labour increases wages that firms pass through to prices, increasing inflation and diminishing capital available for investment. As firms find increasing costs of capital, investment in capital-augmenting technologies is also bound to increase.

The dipping labour share and slowly increasing capital share signifies that capital is yet below the level when it is equally profitable to invest in labour- or capital-augmenting technologies in the US. 

Considering labour-capital complementarity, labour-augmenting technologies (such as the gig economy) should continue garnering firm preferences in the short-term. 

As previously suggested, government should foster existing capital-augmenting technologies’ depreciation (in the linked post I suggest the same for Europe but on labour-augmenting technologies,, considering the final chart of this post) and encourage take-up of big data and self-driving vehicles to accelerate capital-augmenting technologies and the build-up of firm capital. 

It would foster (2000, p.13, case 2, see also 2003 and NBER) natural interest rate increases until capital-augmenting technologies plateau and labour-augmenting technologies take-up. We could infer that it is too early for labour-augmenting technologies such as the gig economy to be profitable in the medium-term.

The circumstance in Europe is, as we may see below, quite different.

Capital destruction and overcapacity

Amongst the possible laureates with this year’s Nobel Prize on Economic Sciences are Wesley Cohen and Daniel Levinthal who developed the concept of absorptive capacity, a fundamental concept to the economics of innovation that defines firms’ ability to embed innovation available in its operating environment.

In the context of the wider discussion between fresh- and saltwater economists, fresh-water economists would argue innovation differently. 
Saltwater economists would consider innovation as a by-product of increasing maket demand that implies more “efficient techniques of production” and leads firms to “place inventions on the market” concurrently leading to business cycles.

Freshwater economists uphold firms decide when to invest in R&D considering competitors  such that the cost of not-innovating may become unbearable and push the firm to invest.

In a gist, saltwater economists focus on aggregate demand conditions whereas freshwater economists focus on firm supply conditions. In that, the putative laureates are clearly freshwater economists in their concern.

If we focus on the freshwater perspective though, some interesting observations come to fruition. For instance, what happens in the medium-term when firms undergo exogenous shocks that destroy capital – do firms (and in that economies) restore capital at the same growth rate or at a faster/slower growth rate?

Consider an exogenous shock (a financial contraction that prevents debt re-financing) that effectively reduces competition. Textbook economics explains  capital is written off and the remaining industry agents now face less competition hence ramp up production and invest to capture freed market portions.

Shliefer, a saltwater economist, forecasts a split in investment between overcapacity and other sectors with investment recovery on the former lagging the latter. Deloitte observes prices should remain subdued in the aftermath of the Recession as firms’ slack implies growing demand is met without further investment.  

Yet the cycle only becomes visible when plotting factor and input intensity of sales – the chart above demonstrates peaks and troughs are so clearly visible that firms’ lagging investment may actually be self-defeating. 

The chart also disproportionately represents the role of demand in determining firms’ investment behaviour (reactive function, the peaks) and firms’ lagged reaction to demand (proactive, the troughs). 

Is the gig economy here to stay?

Reading about the gig economy, I can’t help but think of regulatory and structural change. (Chart data are AMECO rebased to 2000, own calculation).

Economic theory predicts that, given an elasticity of substitution of labour for capital below unit (hence that capital and labour are complementary), a negative shock to investment (such as a financial collapse) implies more unemployment or lower wages as labour becomes less productive.

Since 2009, labour has caught up with the secular trend of higher capital intensity to improve the ratio of labour to capital in the economy. In Japan, the trend inverted in 2002 such that firms increase the labour content of output faster than net capital stock.

The trend was sufficient to invert firms’ preference for capital investment and lower the trend pace of the unemployment rate in the largest OECD economies below that of net capital stock.

The trend predates 2009 as returns on capital dipped below the investment pace between 2005 (US) and 2007 (EU) and has since failed to recover above the level observed in 2000. Understandably, firms prefer labour to capital investment except in Japan where returns to capital increased vis-a-vis capital intensity per worker in 17 of the 18 years plotted.

As firms’ preference steers towards labour, the unemployment rate dips below the pace of net capital accumulation

and pushes returns on labour above returns on capital.

Considering the 2010 IMF’s estimate of the elasticity of labour to capital substitution (.99) for the US and the EU, I apply that value to the index of elasticity for the same year computed by AMECO. Trusting these figures, the elasticity is now above (.04%) unit such that labour and capital are substitutes. The unit value also impacts prices such that higher returns on labour imply lower returns on capital, as seen in the previous chart. 

Furthermore, economic theory predicts that the labour share shall increase when the elasticity is above unit if the capital to labour ratio increases, a phenomenon observable on the top chart’s upward slope since 2009.

Considering the increase in firms’ cost of capital after 2008, the labour income share would decrease if labour and capital were complementary. The following chart seems to confirm the intuition that until 2010 in the US and Japan, and at least until next year in the EU, labour and capital remain(ed) substitutes. 

Acemoglu assumes labour and capital complementarity and labour-augmenting technology to conclude that an exogenous shock to capital cost (a supply shock) would reduce employment, increase wages and lower the interest rate. The shock would also reduce capital intensity and thus increase the labour share until the interest rate is restored.

This is quite visible in the following chart. The dip in labour share between 2007 and 2009 is consistent with supply shock – it was only when interest rates decreased in 2009 (and most significantly in Europe from 2013) that the labour share and wages began increasing. The ECB finds that half the contraction in GDP after 2008 is due to a loan supply shock. The chart also hints at lower policy interest rates as cause of subdued demand for capital, an observation I’d previously made based on firm theory of prices.

The effects on unemployment are visible in the following chart. As predicted by Acemoglu, unemployment was above the pre-recession level until 2017 while wages to capital increased, firms preferring higher remuneration of labour to investing in capital (considering the loan supply shock).

The 2008 recession had financial causes such that firms and households defaulted on loans conceded in the (unfulfilled) expectation of higher productivity. As visible below (and also on the chart comparing net returns to capital with capital per worker) lower returns to capital increased the growth pace of earnings from between 2005 (US) and 2007 (EU). 

Thus, while interest rates remain low, firms shall refrain from resuming capital accumulation. The decision is compounded by the substitutability of labour and capital observed in recent years such that the labour share increases and encourages investment in labour-augmenting technology.

The pressure put on wages vis-a-vis the  elasticity of substitution further encourages, particularly since 2009, investment in labour-augmenting technology: “because wages were growing, and were expected to grow, [so] technical change would try to save on this factor that was becoming more expensive.” (Acemoglu, p.2, op. cit.)

The trend is long-standing, more in Europe than in the US as in Europe wages increase much faster than the labour share, a consequence of lower returns on capital visible on the second chart.

The trend is further reinforced by low unemployment despite an increasing labour share and increasing wages.

Firms are thus prone to continue investing in labour-augmenting technology that increases labour use while improving wages due to labour scarcity. The degree of substitutability compounds the trend, leading firms to explore new ways to save on increasing labour costs – including the gig economy.

Economic theory further suggests capital should accumulate faster when substitutability is significantly above unit. The recent (2017) dip in the unemployment to capital ratio below the 2000 level seems to hint at such an increase while lower returns to capital may withhold faster capital accumulation.

The gig economy relies on labour-augmenting technologies instead of capital-augmenting technologies : extending work schedules instead of buying new vehicles (Uber) or opening more restaurants (Deliveroo), renting rooms at the expense of privacy and comfort instead of building more hotels (Airbnb), even hiring specific skills on a needs-basis instead of permanently (TaskRabbit).

Alternatively, firms could improve capital to increase its complementarity with labour (new mobility solutions, self cooking machines, space-saving interior design and so on) yet that is likely to happen in the current juncture only if capital obsolescence increases (eg a shock that nudges capital replacement such as 5G networks or a policy-induced higher fiscal deductibility of investment as I previously argued or policy mandated pollution limits).

Acemoglu foresees that when labour-capital elasticity is above unit (thus, when these are substitutes) the system does not return to stability. It leads to capital accumulation as visible on the top and second charts and to capital innovation conditional to the capital rate of return. 

Going forward, the value of innovation is expected to increase as interest rates increase. Current policy interest rates below the growth rate make for a negative value of innovation of either capital or labour varieties, with the value of capital innovation further depressed by the (perceived) substitutability of capital and labour. 

As wages increase, so shall the price of labour-intensive goods relative to capital-intensive goods. At this juncture, lower interest rates lower the cost of capital goods, dampening investment in capital-augmenting technologies.

Thus, in conclusion, the gig economy is much the result of capital-labour substitutability, a (loan) supply side shock to the economy and lower interest rates. Growth in capital intensive goods, and hence capital-augmenting innovation is due to remain subdued as long as the combination of higher wages and increasing labour-intensive goods does not put pressure on prices (as it seems to be doing) and thus as long as it withholds higher interest rate (this is visible in the top chart’s upward slope of the EU curve).

Capital-augmenting innovation shall most likely tackle persistently low net returns on capital (as documented on the second chart), an empirically observed determinant of firm entry into investment.*

This is a labour intensive recovery. 

* Thus there is both a demand and a supply effect of the recession – yet policy had focused mostly on demand side effects.

Do higher wages lead to more imports? 

A recurring phenomenon that puzzles Portuguese economists and pundits alike is the increase of imports of foreign goods and services when a profligate Government is instated.

Whereas the puzzle could be easily explained away by Portugal’s relative advantage in mid-tech and low-tech goods (from machinery to artisanal products, where foreign value added is lowest) accompanied by generous spending modernising the country’s infrastructure, the most recent inversion from a current account surplus to a deficit further puzzles as Government spending lags economic growth.

Literature on these matters is scarce as the link is typically established from the current account’s position (following Ricardo?) to the domestic economy, exploring effects known since Harrod, Balassa and Samuelson‘s or Podpiera‘s works; or explaining how wages could affect (or not) export competitiveness.

Could it plausibly be that government profligacy stimulates imports?

The ECB finds government signalling increases private investment at four-fold the rate it increases private consumption. We could say the real exchange rate is rapidly appreciating but wages are relatively stable from 2016. GDP grew faster in recent years when exports increased irrespective of wages although in several quarters a slower pace of wage growth compared with GDP may have induced competitiveness.

Exchange rate appreciation is not what is happening today though as the widest gap between the levels of GDP and wages in the past half decade did not seem to decrease exports – quite the contrary, it is imports that increase.
There are certainly numerous and more complete explanations of the phenomenon, including regulatory changes in export finance and port facilities.

Government spending is known to significantly impact firm investment by crowding out lending to firms and consumers through higher interest rates – and yet, interbank rates decreased faster than Government borrowing costs (not plotted) while lending to the private economy picked up from recent years of deleveraging despite increased public borrowing.

Whereas increased lending may be part of the equation, it certainly leaves plenty to be explained as a contraction in lending occurred concurrently with an upshot in imports.

What could have happened though is a collapse of firm profits as wages increase above productivity. It is known that firms absorb expected higher wage costs by reducing profits. As firms rely on retained earnings to invest, less profits imply Portuguese innovation lags foreign competitors’ so that consumers may prefer foreign innovation while exports grow at a slower pace.

The European Innovation Scoreboard finds Portuguese firms invest significantly in marketing innovation (although increasingly also on product and process innovation) as firms prefer deepening existing product categories to investing in longer-term product differentiation (the score on sales of new-to-market/firm innovations is amongst the lowest in Europe).

Firms would lose competitiveness at home and abroad – and while only a proper statistical regression study could determine the accuracy of this model, statistical aggregates seem to show firm investment was negatively impacted by increasing labour costs above the pace of productivity.

This (plausible) effect is more visible just before 2010 as the price index for domestic investment goods collapsed while the gap between productivity and earnings widened. The index subsequently recovered as firms (plausibly) benefitted from gains in productivity above the pace of gains in earnings, higher profits and more competitive markets. As late as 2017, firms had yet to achieve such demand for investment even as GDP increased by 8pp in-between these years. 

Trump’s tirade du jour

President Trump compared growth in GDP and unemployment to much displeasure and negative criticism of macroeconomists and pundits.

Specifically, the President described the current economic moment positively as the unemployment rate percentage was nominally lower than the percentage rate of GDP growth.

How could unemployment and GDP be linked?

For instance, Rudd and Whelan argue the labour wages portion of GDP is not a good measure of either output gap or inflation, contradicting Woodford’s estimation of the new Philips curve – that assumes ULC and unemployment are related.

Conversely, Tatierska for the Central Bank of Slovakia validates the curve’s prediction and finds a nexus between labour costs and prices via marginal costs that vary proportionately to the cost of labour.

Labour costs increase the elasticity of labour demand for industries where labour costs are a high proportion of total costs such as certain service industries. The elasticity of labour demand is also affected by the rate of substitution between capital and labour – and it has impacted the participation rate, at least in Europe

The effect of either increasing investment or capital on labour intensity of GDP is not immediately visible in German statistics plotted above. Bruegel concludes labour and capital in Germany are in fact complementary.

Assuming a stable savings rate, a declining labour intensity in GDP that would hint at lower consumption even as employment and investment are increase. 

What Trump notices is (perhaps) the low elasticity of labour substitution that enables increasing investment and wages to positively influence GDP. 

The twin increase is not new nor it directly determines growth of GDP – yet it is still good news.

Drivers of the investment cycle

In the follow-up to yesterday’s post on wages and investment, it is interesting to assess what stimulates and impedes firm investment.

Research finds leverage is determinant to firms’ investment decisions when subject to cyclical constraints such as those imposed to European firms between 2011-2016. In other words, market interest rates exogenous to the firm affect firms’ investment decisions conditional on firm indebtness. 

Less explored is the role of policy makers in signalling macroeconomic conditions where firms operate. Whereas uncertainty, in prices or demand, are typically found to affect investment, another less explored signalling channel, the Central Bank interest rate, seems to influence as much. 

Research for the Reserve Bank of Australia concludes that, although firms’ rate of return required to invest is the result of variables other than the Central Bank’s interest rates a signalling channel appears to exert some influence in firms’ decision to invest.

According to Lane and Rosewall’s research for the RBA, a seldom updated hurdle rate prevents firms from recognising improved investment opportunities in lower policy interest rates. 

The complex chart above plots US indicators indexed to 1998 between 2000 and 2010. The chart plots interest rate policy effects (10-year Treasury yield, light blue) and how it eventually impacted corporate bond yields (Baa, purple). The ratio between the two rates (dark red) peaked as the corporate debt lagged the policy interest rate and dipped afterwards until 2005. 

Firms’ capital intensity (orange) behaved accordingly increasing until 2005 and decreasing as the corporate-to-public debt yield ratio began to increase in 2007.

Even as profits (green) dropped from 2007 onwards, firms kept adding debt (dark blue, indexed to 2000), profiting from lower corporate debt rates and ignoring the interstate rate’s steep ascent started in 2006.

Although such superficial research is ineffective in finding what motivated firms to continue borrowing despite increasing yields, investment bank deregulation in 2004 certainly increased market supply of corporate debt. Plausibly, manufacturing corporations increased ownership of non-productive assets (cf chatte below, current assets, purple) and indebtness (dark blue), becoming more vulnerable to financial market events.

Although it likely contributed to higher profits (green) without increasing (productive) capital intensity, the decline in manufacturing employment and capital intensity after 2004 created a more feeble economy with lower aggregate earnings and domestic demand. 

Corporate financialisation plateaued in 2014 even as corporate indebtedness continued a rapid ascent started in 2012 – at least, firms are now investing in productive assets (or so it seems).

Concluding, whereas firms’ investment decisions are typically unrelated to macroeconomic developments, aggregate borrowing and financialisation seem somewhat sensitive to regulatory changes affecting the opportunity-cost of investment.