Forecast of Louisiana Unemployment Insurance Claims. September 2014

Transcription

1 Forecast of Louisiana Unemployment Insurance Claims September 2014

2 Executive Summary This document summarizes the forecasts of initial and continued unemployment insurance (UI) claims for the period September 2014 to August The econometric model for initial UI claims is driven primarily by forecasts for economic conditions, accounting for the time series structure of the data. Continuing claims is modeled as dependent upon past initial claims and current economic conditions, again accounting for seasonal and cyclical patterns in the data. The numbers of initial and continued claims have followed a steady downward trend since the peak of the recession in The state has continued to rebound from the recession and moratorium and is now beginning to see the expected industrial expansions and construction boom start to take shape. A rapid decline in the unemployment rate led to a record low in initial claims over the previous year, as continuing claims also approached a new low. A bright economic outlook for the coming year will lead to a sustained low in initial claims, and additional reductions in continuing claims relative to the previous 12 months. In terms of numerical values, key findings were: Initial weekly UI claims are expected to hover between 2,100 and 3,000 over the 12-month forecast horizon. Continued weekly UI claims are expected to decline overall through the forecast horizon, fluctuating between 17,000 and 22,300 due to seasonal patterns. The forecasted number of continued claims per week averages 19,389 between September 2014 and August 2015, which is 21 percent lower than the average number of continued claims between September 2013 and August 2014, and is lower than the previous annual average low of 20,332 experienced in 2006.

3 Background Over the last ten years, the number of unemployment claims has fluctuated widely due to a string of unusual events including hurricanes, the rebuilding effort and the Great Recession. After Katrina, initial claims spiked for about two months with the lagging effect on continued claims lasting more than half a year. However, as the rebuilding effort picked up speed, a labor shortage began to emerge in the state and the number of initial and continued claims fell to some of the lowest levels in decades. However, as the economy began to return to balance, the effects of the national recession began to impact Louisiana and the number of claims rose significantly with continued claims peaking at just over 68,000 in Since then, the number of initial claims has gradually decreased and continued claims have fallen off significantly aside from normal seasonal upticks. The improvement in economic conditions has been more gradual than many initial forecasts had projected, but the steady pattern of improving economic conditions has brought the number of continued claims back to a level characteristic of a healthy economy. Another important recent change has been the implementation of a reemployment initiative, which is designed to encourage those receiving benefits to find work more quickly. This initiative changed the requirements for collecting benefits to require a person to visit a local office rather than simply file online. This initiative provides more direct assistance to jobseekers and creates a stronger incentive for workers to find new jobs as well as reduce the number of fraudulent claims. The program was fully implemented in April 2013 and early indications provide strong evidence that this resulted in a reduction in the number of individuals claiming both initial and continuing unemployment benefits. However, not enough time has elapsed to provide certainty in estimating the magnitude of this effect given the continuing economic changes impacting the state s economy over the last year. As of September 2014, the economic outlook for Louisiana remains bright. Led by an industrial renaissance fueled by low natural gas prices, the long-term outlook for Louisiana calls for significant growth. In early 2014, the unemployment rate in Louisiana dropped below 5.0%, earlier than anyone expected, and has remained at or below that level through much of Several of the large industrial expansions are underway or expected to begin within the next several months, which will place continued downward pressure on the number of claims. However, with just under 3,000 construction workers counted among those currently claiming unemployment benefits, there is a limit to the short-term benefits that these industrial expansions will offer to the overall number of claims. As the benefits of these large capital investments spread through the economy and the new facilities come on line, the number of continued claims could fall still further in future years. Methodology The model developed to forecast UI claims includes controls that capture the time series properties of the data as well as controls to capture the relationship of economic variables to UI 1

4 claims. Therefore, the model allows anticipated changes in the economy to affect claims, but in a way that is consistent with a lag effect of initial claims on continued claims. The economic variables included in the analysis are the year-over-year difference in US industrial production, a six-month lag in the price of oil, the size of the Louisiana labor force, the state unemployment rate, and a constructed variable to control for the recent recession. Because these variables are used to project UI claims, a forecast for each variable was developed. The consensus forecast from the Federal Reserve Bank of Philadelphia s Livingston Survey was used as a benchmark to forecast US industrial production. The forecast for the price of oil was taken from the futures market. Forecasts for the Louisiana labor force and unemployment rate were based on the state s short-term employment forecast and the population projections that underpin those figures. Finally, a variable was constructed to control for the recent recession using Louisiana employment data to determine the appropriate lagging relationship between the U.S. recession and the recessionary conditions within the state. The equation for initial claims includes the five economic variables discussed above as well as controls for the atypical periods caused by hurricanes Gustav and Isaac, a time trend, and seasonal controls. Modeling the impact of Hurricane Katrina poses special challenges because the magnitude of the storm s impact is large enough to potentially skew the forecast significantly even when estimating the model using over thirty-five years of data. To prevent the atypical period caused by Hurricane Katrina from skewing the results, we opted to exclude this period from the estimation altogether. Another notable change from our previous model is that we added a variable to control for the seasonal uptick in initial claims that occurs the second week of every January as workers who were hired for the holiday season are let go and begin applying for unemployment. To capture the time series properties of the data, one lag of continued claims is included as are six lags of initial claims. This can be formally written as follows:! Initial! = α!" Initial!!! + γ! Continued!!! + β!! + β!! Gustav! + β!! Isaac!!!! + β!! GreatRecession! + β!! DiffIndProd! + β!! OilPrice!!!" + β!! UnempRate! + β!! LaborForce! + β!! Time!!"# + β!! DJan2! + β!" Month!!!!"# The equation for continued claims includes the five economic variables discussed above. The equation also includes controls for the atypical periods caused by hurricanes Gustav and Isaac, a time trend, and seasonal controls. Again, we excluded the period just after Hurricane Katrina from the estimation. We also included a variable to control for the seasonal uptick in initial claims each year during the second week of January. To capture the time series properties of the data, four lags 2

5 of continued claims are included, as are controls for a polynomial distributed lag of initial claims. This polynomial distributed lag places a specific structure on the relationship between the effect of previous initial claims and current continued claims. Because a person can collect regular UI for as many as 26 weeks, previous initial claims as far back as 26 weeks are included in the analysis. However, the effect of those older claims is expected to decay over time as people gradually return to work. Therefore, the polynomial distributed lag terms assume a quadratic decay in the effect of lag initial claims on continued claims. This can be formally written as follows:!"! Continued! = α!" Initial!!! + γ!" Continued!!! + β!! + β!! Gustav!!!!!!! + β!! Isaac! + β!! GreatRecession! + β!! IndProd! + β!! OilPrice!!!" + β!! UnempRate! + β!! LaborForce!!"# + β!! Time! + β!! DJan2 + β!" Month!!!!"# To impose the polynomial distributed lag, or gradual decay, of the effects of initial claims on continued claims, the following relationships are used to constrain the parameters on initial claims and are substituted into the continued claims equation: α! = δ! + δ! i + δ! i! i = 0, 1,, 26 These two equations are solved simultaneously and the forecast is dynamic so that the forecast in each period for initial and continued claims affects the forecast in future periods. 3

6 Results As the improvement in the state s unemployment rate surpassed expectations over the past year, dropping all the way to 4.5% in early 2014, claims for the previous year were better than projected. Weekly initial claims over the 12-month period reached a new historical low of 2,406, and continuing claims approached their historical low as well. With continued growth in construction and industrial activity, the unemployment rate is expected to remain low. As the economy continues to grow, unemployment is projected to remain between %. Despite this leveling-off of the unemployment rate, a strong economy will make it easier for the unemployed to re-enter the workforce and continuing claims will continue to decline. In addition, the full benefits of the recent re-employment initiative may not yet be seen as it could interact with a robust job market in a way that could encourage unemployed workers to find work even more quickly than has been exhibited thus far since the implementation of the program. The updated forecasts for initial and continuing claims over the next year therefore indicate that initial claims will bottom out and remain very close to their current historical lows, while continuing claims will continue to drop, surpassing their previous historical lows. The 12-month forecast of initial claims is plotted in Figure 1 alongside the historical initial claims that were used as a basis for estimating the model. Figure 1: Initial Claims ( ) 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10,000 - Jan- 98 Aug- 98 Mar- 99 Oct- 99 May- 00 Dec- 00 Jul- 01 Feb- 02 Sep- 02 Apr- 03 Nov- 03 Jun- 04 Jan- 05 Aug- 05 Mar- 06 Oct- 06 May- 07 Dec- 07 Jul- 08 Feb- 09 Sep- 09 Apr- 10 Nov- 10 Jun- 11 Jan- 12 Aug- 12 Mar- 13 Oct- 13 May- 14 Dec- 14 Jul- 15 Historical Data 2015 Forecast 4

7 Figure 2 highlights the period since the end of 2009, which shows more clearly the recent historical low in initial claims, and the expectation that it will remain low over the next year. For reference, the September 2013 forecast is included. Initial weekly UI claims are forecast to remain at or near their historical low over the next 12 months, hovering between 2,100 and 3,000, averaging 2,407 weekly. Figure 2: Initial Claims (September 2009-August 2015) 10,000 9,000 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 - Jan- 09 Apr- 09 Jul- 09 Oct- 09 Jan- 10 Apr- 10 Jul- 10 Oct- 10 Jan- 11 Apr- 11 Jul- 11 Oct- 11 Jan- 12 Apr- 12 Jul- 12 Oct- 12 Jan- 13 Apr- 13 Jul- 13 Oct- 13 Jan- 14 Apr- 14 Jul- 14 Oct- 14 Jan- 15 Apr- 15 Jul- 15 Historical Data 2014 Forecast 2015 Forecast 5

8 The 12-month forecast of continued claims is plotted in Figure 3 alongside the historical continued claims that were used as a basis for estimating the model. Figure 3: Continued Claims ( ) 250, , , ,000 50,000 - Jan- 98 Aug- 98 Mar- 99 Oct- 99 May- 00 Dec- 00 Jul- 01 Feb- 02 Sep- 02 Apr- 03 Nov- 03 Jun- 04 Jan- 05 Aug- 05 Mar- 06 Oct- 06 May- 07 Dec- 07 Jul- 08 Feb- 09 Sep- 09 Apr- 10 Nov- 10 Jun- 11 Jan- 12 Aug- 12 Mar- 13 Oct- 13 May- 14 Dec- 14 Jul- 15 Historical Data 2015 Forecast 6

9 Figure 4 highlights the period since the end of 2009, which shows more clearly the recent decline in continuing claims. For reference, the September 2013 forecast is included to highlight how the model has incorporated changes to the economic outlook, and possible evidence for the effect of the re-employment initiative. Continued weekly UI claims are forecast to remain below 22,500 throughout the forecast horizon, averaging just under 19,400 weekly. This forecast represents a new historical low in average weekly continued claims. Also, over the following two to three years, as more industrial projects begin construction and eventually come on line and the effects of the reemployment initiative fully materialize, there is likely to be continued downward pressure on the number of continued claims. Figure 4: Continued Claims (September 2009-August 2015) 70,000 60,000 50,000 40,000 30,000 20,000 10,000 Jan- 09 Apr- 09 Jul- 09 Oct- 09 Jan- 10 Apr- 10 Jul- 10 Oct- 10 Jan- 11 Apr- 11 Jul- 11 Oct- 11 Jan- 12 Apr- 12 Jul- 12 Oct- 12 Jan- 13 Apr- 13 Jul- 13 Oct- 13 Jan- 14 Apr- 14 Jul- 14 Oct- 14 Jan- 15 Apr- 15 Jul- 15 Historical Data 2014 Forecast 2015 Forecast 7

10 Appendix: Detailed Results date i_claims c_claims 9/6/2014 2,349 20,483 9/13/2014 2,316 19,762 9/20/2014 2,267 19,068 9/27/2014 2,272 18,432 10/4/2014 2,208 18,385 10/11/2014 2,170 18,108 10/18/2014 2,145 17,799 10/25/2014 2,110 17,427 11/1/2014 2,075 17,084 11/8/2014 2,496 17,402 11/15/2014 2,485 17,512 11/22/2014 2,428 17,604 11/29/2014 2,487 17,656 12/6/2014 2,648 18,523 12/13/2014 2,639 19,140 12/20/2014 2,634 19,731 12/27/2014 2,666 20,248 1/3/2015 2,676 20,811 1/10/2015 2,222 20,698 1/17/2015 2,348 20,933 1/24/2015 2,520 21,261 1/31/2015 2,548 21,710 2/7/2015 2,419 21,172 2/14/2015 2,348 20,828 2/21/2015 2,269 20,428 2/28/2015 2,165 20,051 3/7/2015 2,171 19,658 3/14/2015 2,145 19,259 3/21/2015 2,107 18,852 3/28/2015 2,088 18,441 4/4/2015 2,530 18,570 4/11/2015 2,477 18,486 4/18/2015 2,380 18,358 4/25/2015 2,411 18,161 5/2/2015 2,391 17,979 5/9/2015 2,263 17,782 5/16/2015 2,299 17,638 5/23/2015 2,344 17,522 5/30/2015 2,350 17,433 6/6/2015 2,622 18,222 6/13/2015 2,740 18,815 6/20/2015 2,824 19,447 6/27/2015 2,956 20,065 7/4/2015 2,754 20,597 7/11/2015 2,731 21,096 7/18/2015 2,744 21,546 7/25/2015 2,692 21,964 8/1/2015 2,653 22,344 8/8/2015 2,151 21,509 8/15/2015 2,147 21,089 8/22/2015 2,192 20,713 8/29/2015 2,112 20,438 8