Despite lower raw material costs – for nearly all of 2016, housing (via $ITB and $XHB) has consistently underperformed the broader S&P 500. Year-to-date, housing related stocks ($XHB) are down 3.4% and homebuilders ($ITB) are down 3.0%. Is this weakness set to continue? The data says ‘yes!’ Demand is rolling over as yields are backing up and while supply is still running rather tight, prices appear to have topped. While the outlook is expected to deteriorate further, fears that such weakness is recessionary, are well exaggerated and premature.
I approximate housing supply with existing homes for sale. Monthly Supply of Houses (MSH) – calculated by the Census Bureau, bottomed in early 2013 and has continued to expand, since. MSH can be thought of as both a measure of existing supply levels, as well as a speedometer of the rate of consumption of existing supply. Regardless of the narrative, supply is elevated and expanding.
It is worth noting that a similar metric – compiled by the National Association of Realtors (NAR), Existing Home Sales: Months Supply, which monitors unsold inventory levels – with much greater volatility than the Census data, tells a slightly different story. According to the NAR, the supply of existing homes is near its lowest level since 2013, currently down ~20% from July 2014 highs. That said, both reports agree that since the beginning of the year, supply has reversed trend and accelerated higher. A rise above 5.0 would confirm the reversal’s staying power and cement the path to much higher inventory in the coming months.
Demand manifests itself through both purchases of existing homes, as well as through construction of new homes. New Home Construction is measured by the number of new housing permits issued and the level of new housing starts – both compiled by the Census Bureau. While builders can’t start construction until a new permit has been issued, a sale can’t be completed without construction having started. Logically then, “permits” must lead starts and starts should correlate more strongly with completed sales. Empirically, the data supports convention, permits do lead starts and they do so by about 1 month.
Since the depths of the financial crisis, we’ve seen monthly demand for new homes more than double to 1337 permits/month in June 2015. However, Summer 2015 appears to have marked a cyclical top in new housing demand. Since then, growth has progressively slowed: (1) monthly permits have come in below the rolling 3-month moving average (3MMA) and they have done so in a series of lower-highs and lower-lows, (2) the 3MMA Divergence Trend – a measure of each month’s divergence from its 3MMA, has set three consecutive lower-highs and (3) the series just reversed back into negative territory in March. In short, permits are rolling over and – according to my housing model, poised to take out 2015’s cyclical low by July 2016.
The rollover in housing permits is largely a consequence of a recent pullback in multi-family units and not necessarily a condemnation of demand for single family units. As the economy exited the depths of the financial crisis, beginning in 2009, both single family and multi-family permits staged an aggressive rebound. Single family permits rose from a -5% consolidated monthly growth rate (CMGR) to a +5% CMGR and multi-family permits soared from a -10% CMGR up to a +5% CMGR. For the next 50 months, single family permits averaged growth of +0.34% CMGR while multi-family permits grew nearly 1.5 times faster at +0.79% CMGR. However, by the Fall of 2015 these trends had diverged and “peak multi-family permitting” had been reached.
Since then, single family permits have maintained steady positive growth while multi-family permits have seen accelerating weakness with a series of four negative, lower-lows and three lower-highs. This divergence doesn’t appear to have run its course, either. March 2016’s multi-family declines were the largest since 2009! Such declines can’t be offset by the steady positive growth in single family permits. With permits leading starts by about a month, March’s dramatic decline should manifest itself in a much lower total starts number, next month.
Starts have tracked very well with their 3MMA, which continues to slope upwards. The 3MMA Divergence Trend has also recently begun to trend higher off of late 2015’s consolidation levels. Viewed in isolation, this data is relatively bullish and suggests positive trending growth for starts should persist. However, when viewed in conjunction with permit data, the picture becomes less optimistic. My housing model predicts April starts will come in around 1045, well within striking distance of 2015’s low.
While the slowdown in multi-family permits has been shown to be the underlying cause of housing’s permitting weakness, the situation does looks contained in the immediate term. Multi-family permits already “caught down” to multi-family starts in March, which suggests multi-family demand forces are balanced in the short-term. Conversely, single family starts looks particularly vulnerable to a sizeable catch-down (~5%) in the following months. A large decline in single-family starts will likely roil housing stocks as market participants incorrectly assume the fall in starts implies further weakness. While further weakness is expected to materialize, a catch-down decline should already be priced in.
Starts/permits tracks backlog consumption, as a second order measure of demand. A positive trending ratio means starts are accelerating faster than new permitting. This represents a drain of demand backlog, where new demand is relatively weak. Conversely, a negative trending ratio means new permitting is accelerating faster than starts. This represents growth in demand backlog, where new demand is relatively strong. A flat trending ratio means permitting and starts are occurring in parallel, where new demand is relatively unchanged. Since 2009, demand for multi-family and single family units has materialized in persistent and divergent trends. While multi-family weakness via starts looks contained in the immediate-term, new demand is expected to continue to be more and more negative.
For single family units, new demand has been almost exactly neutral! For every 100 new permits, roughly 76 new starts occur. That represents dramatic trend stability in an otherwise, exponentially evolving economic environment – nice alliteration, eh! Single family supply and demand have been shown above to ebb and flow, accelerate and decelerate and trend and counter-trend – sometimes dramatically. As such, convention would suggest the stability of this demand consumption speaks to the market’s adeptness at responding to shifts in demand. I don’t buy it! I find it hard to believe US homebuilders are magically and uniquely blessed with the ability to perfectly (slope = 0.00001) match output to incremental demand over the course of seven years. The more likely explanation has to relate to Fed policy and zero rates.
Sure enough, the slope of starts/permits ratio appears to be driven by shifts in the term structure of US interest rates (US 10-year and Fed Funds). During periods of rate declines, the starts/permits ratio fell and when rates rose, so too did the starts/permits ratio. Intuitively, this makes sense. Lower rates stimulate new demand through (1) increased affordability, due to lower embedded mortgage interest expense, (2) broader credit eligibility, due to lower loan-to-value ratios and (3) higher risk tolerance, due to a general “reach for yield.” The inverse is also true when rates rise. Taken one step further, this would suggest that a concept akin to the “non-accelerating inflation rate of unemployment” (NAIRU) may also exist for housing demand. Put differently, there might be a naturally persisting rate of demand for new homes that’s observable during extended periods of unchanging interest rates when the relative push/pull factors of rate declines/rises don’t exist. If this is the case, in the out-years of steady rate periods, homebuilders should become markedly better at managing inventory with new demand. While this idea needs some more investigation, should rates begin to rise, new demand appears very likely to slow.
Since 2009, monthly housing permits correlated most strongly with a 5 month lag on the 30yr conventional mortgage rate (beta = -0.2) and the linear regression on a 1 month lag on the Fed Funds Rate produced the strongest model (r squared = 0.1). Note: Regressions over longer durations (since 1978) showed no correlations with weak statistical significance.
I anticipate inflation will continue to run hotter than expected and we’ll end up seeing a minimum of two additional FFR hikes by the end of 2016. As a consequence of such a policy move, my housing model predicts monthly permits should fall 7.3% to 997/month.
With Fed Funds futures currently pricing in only a 54% chance of a single additional hike during 2016, the market is clearly disinterested in pricing in such risks and housing stocks are no exception. As a result, such a decline in permits would be incremental, above and beyond prevailing trends, which are currently tilted to the downside.
Home prices reflect the market’s balance of supply and demand. Both the 20-City and National Case-Shiller Home Price Indices look to have put in cyclical tops in November/December 2015. Since the start of 2016, both indices have accelerated lower. These trends in pricing support the above narrative of a housing market experiencing stagnant demand and expanding supply.
THE WORLD ISN’T FALLING APART
While it seems the frenzy over fears of a looming US recession appear to have abated – at least for the moment, “recessionistas” point to a slowdown in domestic housing as a leading indicator. As of 1Q2016, housing accounted for 3.6% of GDP. While that figure has continued trending ever higher, it’s still nearly half of the 2006 peak of 6.6%. While the results of my analysis suggest there’s still further downside for the domestic housing market, the US has never entered recession while housing made positive incremental contributions to changes in real GDP. This data would suggest housing is a strongly lagging indicator of recession. Put differently, regardless of weakness elsewhere in the US economy, until housing rolls over, recession is highly unlikely.
Based on the data, recession fears should have been highest prior to 1Q2014. Since then, housing’s contribution to changes in real GDP has turned positive and continues to accelerate. While housing is due for a slowdown, unless we see a dramatic and sustained reversal in housing’s GDP trends, recession fears due to housing concerns appear exaggerated.
In summary, we’ve seen that current demand via permits, is rolling over due to a massive pull-back in multi-family unit investment. While single family unit demand has been steadily positive, starts are due for a catch-down of nearly 5% in the coming month and it appears most of the resilience in single family permitting is due to the persistence of zero interest rates. As inflation heats up and Fed policy turns more hawkish, we should start to see weakness emerge from the single family side, as well. While the NAR data claims supply is tight, its recent trend reversal supports Census data that says supply is growing at an accelerating pace. Moreover, tops in both the national and 20-city Case-Shiller home price indices suggest a supply glut is materializing. None of this bodes well for the future of housing. Even the most optimistic reading of this analysis would only suggest housing has more of the same ahead of it, which isn’t good. Ultimately, though, the data isn’t calling for a collapse nor is it signaling recession. I plan to continue shorting the sector through the next batch of permit data and on further positive momentum in inflation.