Although the economy entered the homestretch of 2019 without much momentum, our revised forecast is that growth will stabilize and we will continue to muddle through. There is, however, an elevated level of uncertainty in the current situation, and that uncertainty has been increasing during the first month of Q4. Caution is warranted.
The advance data for the third quarter were disappointing overall, with a mixture of positives and negatives below the top line. That top line – growth in real output – came in at 1.9% compared to our August forecast for growth of 2.3%. That difference was accounted for twice over by a dismal performance from business investment, which showed significant declines in purchases of both equipment and structures. Households did much better. Consumption came in at 2.9%. This was close to our expectation, and much more sustainable than the Q2 4.6% rate. Housing (finally) had a good quarter, after a year and a half of negative results. Government spending was also strong. We think this brings us close to the end of the surge from the budget deal of early last year. Finally, as we expected the trade balance was about unchanged from Q2. However, this was the result of both export and import growth that was significantly below our forecast.
Overall the Q3 numbers contain about an equal mix of pluses and minuses, netting to a roughly neutral result.
Taking a longer view, a year ago we were clearly too optimistic in our outlook. Growth over the past four quarters (at 2%) has been a full percent below our year ago forecast. Our main error was our expectation that business investment would continue to show growth exceeding 5%. Instead it has grown only 1.2% over the past four quarters. Part of this can be blamed on data revisions, but the prime mover is the trade war escalation of the past year. The story here is that the erratic imposition of tariffs by the Trump administration (and the predictable retaliation by those affected) has disrupted trade flows and with them global supply chains. The resulting uncertainty has then hit businesses willingness to invest in new capacity. A year ago we saw this as a risk, but did not factor it into our baseline forecast.
Overall, the NIPA data suggest a cautious outlook for the next year. Growth in the 3rd quarter was adequate, but not broadly based. It rested on household and government spending, with negative contributions from business investment and from international trade. The household contribution should be sustainable if consumer confidence holds together. Government spending growth seems likely to slow. For overall growth to hold at its current rate some improvement in investment and trade will be required.
Monthly data released during October have been an even mix of mildly encouraging and mildly disappointing.
Indicators that relate to the manufacturing sector fall in the latter camp. Industrial production for September was negative, both month-over-month and relative to a year ago. The ISM manufacturing index rose slightly in October, but still registered its third month of contraction (a reading below 50).
Data ouside of manufacturing are generally positive. Income growth has been good. Consumption spending has been good. Consumer sentiment is holding at high levels. The ISM non-manufacturing index for October rose by 2.1 points to a reading of 54.7. Housing permits were strong in both August and September. Housing starts fell off in September from August’s best-in-a-decade rate, but still show growth over the past six months.
Finally, the labor market report for October was quite good on both sides. Payroll employment from the establishment survey was held back by about 60 thousand due to the GM strike and to temporary workers doing preparation for the Census who have completed (for now) their work. Even so the employment total increased by 128 thousand. Private sector services jobs were up 157 thousand with solid increases in the professional and business, health care, and leisure categories. In addition the employment increases for August and September were revised upward by a total of 95 thousand jobs.
In the household survey the labor force was estimated to have increased by 325 thousand. This was in excess of population growth, raising the participation rate by a tick to 63.3%. A year ago the pr was 62.9%. While seemingly a small difference it amounts to over a million extra workers. Even more if you factor in demographic pressures that are holding down labor force growth by perhaps 180 thousand per year. The household survey estimate of October employment rose by 241 thousand. While solid this was short of the labor force growth, resulting in the unemployment rate rising a tick to 3.6%.
Overall the new data continue to depict a schizophrenic economy. Trade sensitive goods production is bordering on recession. The (larger) rest of the economy is doing reasonably well. It is not clear this division can continue, but so far so good.
As can be seen in the chart, our model remains relatively sanguine, although less so than in our outlook from three months ago. In our new forecast real output growth averages 2.0% over the next year, which is close to 0.4% below our August expectation. Farther out the difference is smaller, shrinking to -0.2%. In the short-run, we expect the current quarter to be modestly better than Q3, with growth of 2.2%. This comes from growth in business investment that is slow, but an improvement on the decrease in Q3.
We think this forecast is “sanguine” in two regards. First, it has a continuation of the expansion, already into a record eleventh year, into the indefinite future. Second, it reflects somewhat better balance in the economy. As mentioned above, there is at least some growth in business investment. There is also continuation of the Q3 growth in the housing sector. Balancing this is somewhat weaker growth in consumption and in government spending. For consumption this brings spending in line with income. For government it moves in that direction, but the line remains a long way off. In the labor market (Figure 3) the average monthly increase in employment falls off due to demographic pressures on the labor force. There will, however be some dramatic swings from month to month arising from temporary federal government hiring related to the census. Based on the pattern in 2010, we expect this to peak in Q2 and to be reversed in Q3. During the third quarter it is possible that we could see a month in which total employment shrinks for the first time since the Great Recession. In the face of this we expect unemployment to remain at about its current level, with only slight upward movement later in the forecast.
Elsewhere in the economy our forecast assumes that the Federal Reserve will maintain the status quo after its recent third rate cut.
In summary, our model continues to think the economy can continue to muddle through for a while longer. The growth we expect is basically the “new normal” that has characterized most of this expansion – driven by consumer spending, along with weak, but positive, growth in investment, housing, and from government.
A model-based forecasting procedure like ours is problematic in an environment characterized by a high level of uncertainty as at present. In general forecasting models extrapolate on the recent past using reaction patterns drawn from historical experience. This framework produces forecasts that are inherently more stable from quarter to quarter than reality.
The forecaster plays a role primarily by imposing assumptions about elements relevant to the current situation that are outside the scope of the model. Our standard operating procedure in this regard is to configure our assumptions to roughly match longer-run historical experience. Thus, for instance, in our baseline forecast we assume that the stock market variable in our model will rise at a steady rate roughly in line with its long-run trend. Or, for our exchange rate variable we assume that it will be constant at its current level. In the short-run neither of these assumptions (and others that we make) are likely to be accurate, but to assume something else would simply be guesswork. That said, this procedure clearly imparts a status quo bias to our forecasts.
In the present situation, in which erratic trade policy is having an unprecedented impact on the economy, our model does badly. To begin with, there is nothing in the historical record of the past half-century to draw on, meaning that our model essentially extrapolates a future in which trade policy is irrelevant. Even if our model did include trade policy effect, we would still need to make an assumption about the future path of that policy. With our sop we would assume that the state of policy would remain about unchanged, again making it largely irrelevant to our forecast. We don’t think trade policy is in fact going to be irrelevant. But we don’t know how to put it into our model. And even if we did, we don’t have any firm belief about how policy will change over the next few quarters.
A year ago this procedure gave us a forecast that proved optimistic when the trade war significantly worsened. Currently (meaning the past couple of weeks) the market seems to believe that the trade situation could improve. If that in fact proves to be the case, our forecast could be too conservative.
On the other hand, we think there is about an even chance that the recent appearance of progress could reverse, with a downside impact on the economy relative to our outlook.
On the downside the largest risk, to which trade policy could contribute, is a significant retrenchment by consumers. At present we have an historically strong labor market and solid increases in income leading to high consumer confidence. But the labor market is a lagging indicator, and consumer confidence can change in a hurry. If it does all bets are off.
But for now, we think the most likely outcome is a continuation of modest growth.