Data Science Software

I have been using the Anaconda release of Python 3.x for a few months. The package updater makes it easy for the maintenance part of my computer. I have been playing around with integrating Python and R and tonight installed the package r-essentials with the command:

conda install -c r r-essentials

Obviously, this post is for myself so I do not forget that command. This should keep me busy for awhile.

February 2016 Employment

The February 2016 employment rates were released on Friday. It was a very good report. The participation rate increased to 62.7%. That is now right at the twelve-month moving average and an increase from January’s reading of 62.5%.

There are two parts of the report I watch closely. The narrative of the recovery has been that older people are staying employed longer or going back into the workforce because their savings isn’t enough to fund their retirement in this low interest rate environment. February was not a kind month to this story line. The participation rate for men and women between 16 and 24 years of age jumped to 53.7% from January’s rate of 52.9%. This was above February 2015’s rate of 53.6%.

Meanwhile, men over the age of 65 increased their participation rate to 23.6% from January’s rate of 22.9%. Women over the age of 65 maintained their participation rate of 15.6%.

Our calendar age may be increasing, but our activity level isn’t decreasing at the same pace as advances in health care can keep us active longer. I hear negative stories about how bad the savings rate was for the oldest section of the baby boom generation. Yet I do not consider it to be a bad thing they are staying in the workforce since it is a positive contribution they are making.

I also hear negative stories about how this increase in the participation rate was accomplished by low-wage jobs increasing at a quick pace. I think these are the same people who complained a year ago when the only jobs being created were very high skill jobs. The current broadening of the employment situation is a very good sign for the strength in the economy and should continue to push consumer based consumption higher in the coming months. I have resigned myself to the conclusion that some people are always critics and they are likely the ones who are not contributing to society.

There were a couple of other indicators that were released on Friday in addition to the employment report. The combination of these indicators caused the Atlanta Fed to raise its estimate of first quarter GDP to 2.2%. This is well above the fourth quarter second estimate of 1.0%. It is well past time to recognize there is a significant recovery happening in the US economy and we are leading the world in the recovery.

CPI Predictor

After last week’s post, I worked on the predictor for the direction of CPI. I’ll start with the construction of the predictor and then move on to the construction of the test.

The construction of the predictor uses the eight components of the CPI as factors. Two values are generated for each factor — the current year-over-year change in the price index and the previous year’s year-over-year change in the price index. These two numbers are compared with those showing an increase being included in the result.

Let’s use Medical Care as an example. From January 2014 to January 2015, the price index changed from 429.621 to 440.969. Then from January 2015 to January 2016, the price index changed from 440.969 to 454.175. This is an increase of 2.64% to 2.99%. Because the year-over-year price index showed a year-over-year increase in the percentage rate of change, the weight of the factor is included in the January 2016 result. Medical Care for January 2016 has a weight of 8.375.

Five of the eight factors showed a year-over-year increase in the percentage rate of change and their combined weights are 39.614. This number has been increasing since June of 2015 when the weighting result was 0.0.

The weighted year-over-year change of those factors showing a year-over-year percentage increase is 0.79%. This modest amount is nowhere near the Federal Reserve’s 2% target, but it is an increase from December’s reading of -1.32% (the Transportation factor had an unusual impact on the final result).

The test that I created was to compare the correlation of the result to the subsequent month’s annualized ten-year CPI percentage change. The hypothesis was an increase or decrease in the size of the weights in the result would cause an increase or decrease in the following month’s ten-year annualized CPI price change (this is purposefully worded to show the expectation of a positive correlation).

The correlation showed a positive result of 0.6655. Constructing the test statistic resulted in 4.4576. The 95% confidence interval of a two-tailed T test gave a critical T value of 2.0595. Since the test statistic is greater than the critical value, the null hypothesis can be rejected and there is sufficient evidence the predictor does lead to a change in the ten-year price change.

Using this data and calculating the test could not have been done without the class I am teaching this quarter — Introduction to Business Statistics. I know the students are wondering why they need to take this on the journey to their Business degree, but the logic of constructing a hypothesis and test will serve them later.

January 2016 CPI

The January 2016 CPI numbers were released on Thursday. Headline inflation over the past ten years has averaged 1.947% per year. This is down from December’s number of 2.019%. In January 2015, the number was 2.255%. While it has not been a straight road down, there is a clear downward trend in inflation.

With the January 2016 release of the CPI numbers, we also received a change in the weighting of the components within the headline number. Next I’m going to walk through the important components and their changes.

The largest component is Housing. In January 2012, the weighting was 41.020 and it has been moving upward for four years to 42.235 in January 2016. Meanwhile, the ten-year inflation rate for Housing has been decreasing from 2.433% in January 2012 to 2.021% in January 2016.

Transportation is next and its weighting has decreased each year over the past four from 16.875 to 15.259. The ten-year inflation rate for Transportation has rapidly decreased from 4.186% in January 2012 to 0.811% in January 2016. Within the Transportation component is Motor Fuels. While I do not track this sub-component, I do track the Core Measure of Energy, which is a combination of Motor Fuels and the Fuels sub-component of Housing. The weighting of Energy has decreased from 9.679 in January 2012 to 6.816 in January 2016. The ten-year inflation rate for Energy has decreased from 11.212% in January 2012 to -0.492% in January 2016.

The third component I want to mention is Medical Care. The weighting has moved up all four years from 7.061 to 8.375. The ten-year inflation rate has moderated from 4.594% to 3.784%. This number is still above the headline number and combined with the change in weighting will continue to make this number pull inflation higher. Also, notice Medical Care had a lower weighting than Energy in January 2012 (7.061 vs. 9.679) and how that has reversed in January 2016 (8.375 vs. 6.816). If I were to make a claim, maybe the gasoline savings for consumers is now being spent on Medical Care. CPI numbers can’t verify that since they report the rate of change in price levels and we would need expenditures in these areas to justify that claim.

In looking forward, there is a leading indicator of what to expect from upcoming CPI figures and it has been moving higher for several months now indicating the headline CPI number should begin to move higher. For several months now, I have a to-do item to look at this indicator and determine if it is useful. A simple feel test tells me it has a high emotional element but a low accuracy factor. If I ever get to examine it, I’m certain there will be a post about it in the future.

Retail Sales from Friday

Retail Sales was released on Friday. The release impacts real consumer spending, a large component of GDP. The consumer has been driving the increase in GDP over the past few quarters. Prior to this release, the consumer spending growth was 2.0 percent. With a GDP forecast of 0.8 percent, it is easy to see that consumer spending is driving the growth in GDP.

The release on Friday dropped the consumer spending growth rate to 1.7 percent. Because of its out-sized influence, the fourth quarter GDP estimate dropped to 0.6 percent. I am starting to become alarmed, but not because of the drop in the estimate. The blue chip consensus saw an increase in the estimate for GDP as the high-end of the estimates moved up to 2.5 percent. The divergence that occurred with the Friday release of GDP Now is startling. What are other economists seeing that would result in their estimate increasing?

The next estimate will be released on Wednesday.

GDP Update

The GDPNow estimate at the Atlanta Fed has experienced a wild week. Let’s go through the changes so far. The estimate had been stable for a couple of weeks through the end of 2015 as there was no activity. We began the week with a fourth quarter estimate of 1.2%.

On Monday the ISM Manufacturing Index was released and it was dismal. When I go through the numbers remember this release was just for the month of December.

  • The estimate on PCE Goods went from 3.0% to 2.5%
  • The estimate on Fixed Investment went from -0.4% to -2.1%
  • The estimate on Residential Investment went from +1.3% to -2.9%

Overall, the GDPNow estimate fell to 0.7%. This is a significant cut for a number that impacts only one of the three months in the quarter.

I heard exactly one reaction and it was flawed. On Tuesday the minutes from the December meeting of the Federal Reserve were released. This was the meeting where the Fed decided to raise rates. After the minutes, everyone was quick to note the vote to raise rates was barely in favor. The analyst I heard mentioned that if the Fed knew the GDPNow number would drop to 0.7%, it was likely they would have held off on raising rates.

I think that is incorrect. The Fed is responsible for stable prices and increasing employment. They are not responsible for growth of the economy. Growth in the economy has two precursors — productivity growth and population growth. With CPI or the PCE deflator well under 2% and employment continuing to increase, the Fed is achieving both targets and beginning to normalize rates is the correct response given their mandate.

Yesterday, figures on International Trade were released. This impacted GDPNow by raising the estimate of Q4 to 1.0%. While exports remained at 0.2%, imports went from 2.5% to 1.0% and since Net Exports is Exports minus Imports, the decrease in imports is a net positive for GDP.

Tomorrow is jobs day in the US as employment figures are released for December. This will not impact GDPNow. Later in the morning, figures on Wholesale Trade will be released and that will have an impact. Once we get those numbers, economic statistics affecting GDPNow will be absent for one week.

The current estimate from Blue Chip Forecasts that the Atlanta Fed uses is constructed by taking the top ten forecasts and the bottom ten forecasts. The range for Q4 is 1.5% to 2.6%. This is well above the GDPNow estimate of 1.0%. We are three weeks away from the official first estimate which leaves plenty of time for revisions. I hope to have time to follow this as it appears this might influence Fed actions. It shouldn’t, but if someone expects it to enough to make the comment publicly, there could be some interesting commentary for the January Fed meeting.

GDP and Recession Risk

As we face the first increase in the targeted Federal Funds rate in nine years, I hear some analysts complain the Fed is behind the curve and others say the economy is likely to move into recession because it cannot handle the increase in interest rates. No matter which side you are on, there are already some indicators we can follow to see where the economy is headed and what the probability of recession is.

Two of these indicators we are already following. The first is the expected inflation rate for the next five years. This indicator is not based on actual data, but rather is a number tracked by the bond market. Thus it is an indicator of where market participants are betting the inflation rate will likely be. As of December 10th, that number is 1.20%. It was 1.30% at the end of October 2015. It was 1.27% on December 10, 2014. That is a narrow range and stable over the past year. It is also nowhere near the Fed’s stated target of a 2.00% long-term average for inflation. The conclusion we should come to is the Fed does not need to raise rates since the market is saying inflation, one of the key Fed guiding measures, will not be a problem for the next several years.

The next indicator to observe is the two-ten spread. This measure is the difference between the yield on the ten-year U.S. Treasury bond and the yield on the two-year U.S. Treasury bond. When this indicator turns negative, there is a high likelihood of a recession starting within two quarters. As of December 10th, the spread is 1.29%. As of October 31st, it was 1.41%. As of December 10, 2014, it was 1.59%. It has not been a straight line down over the past year, but a thirty basis point move over the past year is something to follow. For now, this indicator is well above zero. I do not expect to get concerned until it drops below 50 basis points. I think the financial media will get interested if it drops below 100 basis points.

The previous two indicators I have been following on the Analytical Road at blogspot. The next two I have just recently been reading about and both are hosted at the Federal Reserve of Atlanta. One uses historical GDP to predict the likelihood the U.S. is currently in a recession. This is useful since the NBER is the official body which can declare a recession but can sometimes have multi-year delays in their decision. The second uses current data and estimates for an in-period forecast of quarterly GDP. We will start with a look at that indicator.

GDPNow is updated at least once per week. On December 4th, the forecast for the fourth quarter was 1.5%. On December 11th, it moved up to 1.9%. The change over the week comes from the increase in real retail sales and the decrease in wholesale trade and retail inventories. At 1.9%, it won’t raise any concern even though that is a low number. A 1.5% number for the fourth quarter would raise concern. Using the current estimate of 1.9%, it appears the economy is not outgrowing its long-term average. There aren’t too many more financial numbers before the end of the quarter and I suspect the estimate will start to decrease volatility and the focus will turn to the first quarter of 2016. For now, this indicator seems to suggest no change is needed.

The other indicator from the Atlanta Fed is called the GDP-Based Recession Index. This indicator uses historical GDP and a large number of probability estimators to estimate the likelihood the U.S. economy entered a recession in the previous quarter. This indicator was last updated on August 10, 2015 and reads 13.3%. This is significantly below the level of 67% which is the trigger to indicate a recession has begun.

Of these four indicators, the forward five-year inflation rate and the two-ten spread are collected daily. GDPNow is updated weekly. The GDP-Based Recession Index is updated quarterly. I have thought about putting these together into a single page, but there is another indicator we need to watch because we have an extraordinary market with the Fed’s zero interest rate policy. This measure also comes from the Federal Reserve of Atlanta and is called the Wu-Xia Shadow Federal Funds Rate. When the Fed moved rates to zero, there was academic talk about the lower bound of zero. While some European economies and the ECB have moved their rates below zero, the Fed has held rates at the zero bound. Researchers Cynthia Wu and Fan Dora Xia used some clever data to calculate what the Federal Funds rate should have been. The shadow rate reached a low in May 2014 at -2.99%. As a saver, I am very pleased the Fed did not lower rates to that level. The rate as of November 2015 is -0.004%. Since that May 2014 low, this shadow rate has been moving very quickly higher. It seems reasonable the Fed would want to stay near this rate as this is what the market and economy indicate is necessary. In that case, a 25 basis point move this week would not be unreasonable.

The result of this walk through our data is to conclude that a rate increase is needed by one measure and not necessary by others. Our two daily indicators are subject to the emotional volatility of the market and based more on what market participants expect rather than what has occurred. The next week will be interesting to see how these indicators react. I am assuming the Fed does increase rates but will also say future increases are data dependent. I am not expecting another date to be explicitly mentioned. I would expect that to be the unwelcome market moving event. Once we move off the zero based policy, the shadow rate will go away. What we will be left with is two market based indicators and two GDP based indicators as a set to indicate what the Fed is likely to do. I will work on making a cohesive case for these over the next few months.