TRACI PUDELSKI and ERIK EKEBERG
College of Business
Western Carolina University
Abstract
U.S.
employment is expected to be between 5.6 percent and 7.4 percent during
the next seven years. This forecast is based on the Keynesian aggregate
expenditure model using unemployment rates, investment spending, and interest
rates as the explanatory variables. The relation between the unemployment
rate and the explanatory variables is weak. The R-square indicated that
only 13.32 percent of the variation in the unemployment rate is explained
by the other three variables (unemployment, investment spending, and interest
rates). This low R-square could be a result of the three variables being
lagged seven years. According to the forecast the workforce is going to
increase considerably over the next seven years (JEL: E22, E24).
Part 1. Introduction
This
paper forecasts the monthly unemployment rate for the years between and
including 2000 and 2006.The data
period used is from January 1980 through December 1999.The
data includes the high rise in interest rates at the beginning of the recession
in 1981 up until the end of the available data in 1999.The
explanatory variables are previous unemployment rates, private domestic
investment, and interest rates.The
forecast is for seven years and is based on the Keynesian aggregate expenditure
model and the relationships between investment spending, interest rates,
and unemployment.This model is
easy to understand and explains the relationships between the variables
accurately.
Using
the sample data to predict unemployment into the future allows the government
and prospective employers to plan for the surplus and/or deficit of capable
employees.This forecast will assist
in deciding on future regulations, projects, expansions, cutbacks, tax
rates, etc.A forecast of unemployment
rates will also be valuable for current and future employees.Current
employees will have some idea of their level of job security and future
employees will have some idea of the difficulty in entering the workforce.
Because the forecast horizon covers seven years, the forecast could have possible inaccuracies.The sample data used, which includes a significant recession, could also be a factor of inconsistencies in the results.Taking into account any possible discrepancies, the rest of the paper is divided into five other sections starting with: Part 2. explains the data being used to predict the unemployment rate; Part 3. presents the economic theory and relationships between the variables in order to predict the unemployment rate; Part 4.displays the empirical results and discusses the validity of those results; Part 5.evaluates the importance of the forecast for the economy; Part 6. and discusses conclusions and possible explanations.
All
of the variables are taken from the Federal Reserve Bank of St. Louis Federal
Reserve Economic Data (FRED).The
measures for Real Gross Private Domestic Investment, the Civilian Unemployment
Rate, and the 30-Year Constant Maturity Rate are FRED variables GPDIC1,
UNRATE, and GS30, respectively.
Real Gross Private Domestic Investment (GPDIC1) is given in billions of chained 1996 dollars at seasonally adjusted annual rates.The primary source of the data for this variable is the U.S. Department of Commerce, Bureau of Economic Analysis.As a result of this variable (GPDIC1) being observed quarterly and the other two variables (UNRATE and GS30) being observed monthly, data manipulation was conducted on GPDIC1 to convert it into monthly observations.In order to make the time frame consistent with the other two variables, each quarterly observation of the GPDIC1 variable was duplicated for the
following
two consecutive months of each quarter corresponding to each individual
observation (e.g., the observation in 1980.1 of 711.7 was duplicated to
1980.2 and 1980.3).
The
Civilian Unemployment Rate is seasonally adjusted and is given as a percentage
of the total population.The primary
source of the data for this variable (UNRATE) is the U.S. Department of
Labor, Bureau of Labor Statistics.The
other variable (GS30) is given as a percentage and is based on averages
of business days.The primary source
of this variable is the Federal Reserve Board of Governors - H.15 Release.Neither
of these variables required any type of data manipulation.
The
forecast is for seven years into the future.Retrieving
the information and preparing the data in order to perform a regression
and calculate a forecast for the next seven years was relatively easy.However,
forecasting this far into the future will obviously include inaccuracies.Using
seven years to forecast means that a larger time period of previous data
has to be used.Using a larger
time frame of sample data increases the probability that certain observations
in the sample data will not be typical in the forecasted time period.For
example, the following graph (Figure 1) shows that the unemployment rate
was forecasted to increase by nearly 1.5% in the first month of 2000.
Where Y is real GDP, C is consumption spending, I is investment spending, G is government spending, and X or (EX – IM) is net exports.Net export (X) equals exports (EX) minus imports (IM).Simplifying the Keynesian model by applying the assumption that certain factors can be held constant, we can rewrite the model as:
Y = f (I)
The next step is to introduce the aggregate production function, which is written as:
Y = f (L, K)
Where L is labor and K is capital.This equation can be rewritten as:
Y = f (L)
Stated this way, GDP is written as a function of labor, which allows the forecast to conclude that GDP is also a function of the employment rate, and inversely a function of the unemployment rate.The relationship between GDP (Y) and the unemployment rate (u) can be rewritten as:
u = f (Y)
Last is the relationship between investment spending and interest rates (r):
I = f (r)
Using this principle there is a presumed relationship between interest rates, investment spending, and unemployment rates.Combining these principles together, the hybrid model of unemployment can be written as:
u = f (u, I, r)
After lagging the right hand side of the equation seven years (84 months), the forecasting equation is written as:
ut =
A0 + A1 (ut-84) + A2 (It-84)
+ A3 (rt-84)(1)
The
constants A0, A1, A2, and A3
in the equation are estimated coefficients, which are calculated
in the regression and reported in Table 1 on the following page.The
first coefficient is an intercept and the other three are estimated coefficients
assigned to the variables of unemployment rates, investment spending, and
interest rates.
Part
4.A Seven Year Forecast of the
Unemployment Rates
The unemployment function, equation 1, was estimated with 1980.1-1999.12 monthly data.The least squares regression method was used.The regression estimate is:
ut =
1.247361 + 0.009416 (ut-84) + 0.002834 (It-84) +
0.210681 (rt-84)
Regression Estimate
of Unemployment Forecasting Equation 1: 1980.1-1999.12 |
||
Explanatory
variable
|
Estimated
coefficient
|
t-ratio
probability level
|
Intercept
|
1.247360664
|
0.791416
|
Ut-84
|
0.009415685
|
0.112586
|
It-84
|
0.002833896
|
2.479976
|
Rt-84
|
0.210680676
|
4.806573
|
R2
= 0.133821389
|
F
(zero slopes) = 7.827812
|
Prob
F = 6.18793x10-05
|
The R-square of the estimate is 0.133821 as shown in Table 1, indicating that approximately 13.4 percent of the variation in unemployment is explained by variation of the variables: past unemployment, past real private domestic investment spending, and 30-year T-bill interest rates.This is a weak relationship and actually does not explain the forecasted variable very thoroughly.The F-statistic is 7.827812 (Table 1), with a probability of 0.0000682.This is very good and indicates that the null hypothesis of zero slopes is strongly rejected.Although the T-statistic for ut-84 is outside the range of the absolute value of two, it was left in the forecast equation because it is the lagged dependent variable.Investment spending and interest rates were left in because the absolute value of the T-statistic was slightly above two and nearly five, respectively.Using the intercept and coefficients of the variables, the equation can be applied to forecast the unemployment rate for a random month:
u2001.1
= 1.247361 + 0.009416(6.8) + 0.002834(1057.3) + 0.210681(6.29) = 5.631078
Forecasts of U.S.
Unemployment Rate 2000.1-2006.12 (Given as a percentage) |
|
Quarter
|
Forecast |
2000.1
|
5.59674
|
2000.2
|
5.542186
|
2000.3
|
5.484361
|
2000.4
|
5.497857
|
2000.5
|
5.512605
|
2000.6
|
5.488489
|
2000.7
|
5.441406
|
2000.8
|
5.375153
|
2000.9
|
5.306794
|
2000.10
|
5.441045
|
2000.11
|
5.496046
|
2000.12
|
5.503532
|
2001.1
|
5.631078
|
2001.2
|
5.673214
|
2001.3
|
5.760759
|
2001.4
|
6.009103
|
2001.5
|
6.035774
|
2001.6
|
6.033667
|
2001.7
|
6.024262
|
2001.8
|
6.004359
|
2001.9
|
6.049767
|
2001.10
|
6.235298
|
2001.11
|
6.26291
|
2001.12
|
6.217725
|
2002.1
|
6.248178
|
2002.2
|
6.195731
|
2002.3
|
6.162022
|
2002.4
|
6.050755
|
2002.5
|
5.962493
|
2002.6
|
5.882434
|
2002.7
|
5.888338
|
2002.8
|
5.917833
|
2002.9
|
5.851581
|
2002.10
|
5.907089
|
2002.11
|
5.884855
|
2002.12
|
5.842719
|
2003.1
|
5.89795
|
2003.2
|
5.936097
|
2003.3
|
6.011942
|
2003.4
|
6.225129
|
2003.5
|
6.255566
|
2003.6
|
6.280129
|
2003.7
|
6.411724
|
2003.8
|
6.367928
|
2003.9
|
6.408899
|
2003.10
|
6.369068
|
2003.11
|
6.301426
|
2003.12
|
6.316174
|
2004.1
|
6.495518
|
2004.2
|
6.466023
|
2004.3
|
6.515645
|
2004.4
|
6.739049
|
2004.5
|
6.706505
|
2004.6
|
6.671631
|
2004.7
|
6.62489
|
2004.8
|
6.639637
|
2004.9
|
6.623724
|
2004.10
|
6.663394
|
2004.11
|
6.616102
|
2004.12
|
6.591762
|
2005.1
|
6.855944
|
2005.2
|
6.871857
|
2005.3
|
6.885439
|
2005.4
|
6.823207
|
2005.5
|
6.826255
|
2005.6
|
6.77874
|
2005.7
|
6.882218
|
2005.8
|
6.852723
|
2005.9
|
6.781092
|
2005.10
|
6.862357
|
2005.11
|
6.911979
|
2005.12
|
6.87195
|
2006.1
|
6.932603
|
2006.2
|
6.977787
|
2006.3
|
7.020147
|
2006.4
|
6.990963
|
2006.5
|
7.044798
|
2006.6
|
7.094196
|
2006.7
|
7.228357
|
2006.8
|
7.246376
|
2006.9
|
7.246376
|
2006.10
|
7.381253
|
2006.11
|
7.358078
|
2006.12
|
7.400215
|
Due to higher investment spending, this forecast predicts the unemployment rate to steadily increase over the next seven years, as shown in Table 2.As investment spending increases, interest rates will eventually rise.As interest rates increase, the price of consumer goods will rise as a result of inflation.Unemployment rates will start increasing due to lower margins.Those unemployment rates are not favorable for the economy and predict a downturn, or recession, that could possibly produce layoffs.
Part 5. Utilization Declines and the Worker is Unemployed
Many groups and individuals can use unemployment rates for estimates and planning future events and activities.This paper forecasts unemployment to steadily increase over the next seven years, reaching a high of 7.4 percent.This shows a drastic increase from the low unemployment levels that the country now enjoys.With increasing unemployment rates, more people will lose their jobs and income. On the other hand, there will be more employees in the workforce from which businesses can choose.State and federal governments will have to focus more of the budget on unemployment benefits.The higher unemployment rate means more people will be competing for fewer jobs and markets will weaken.