Our formula for projecting United States unemployment rates mirrors the formula used by the US Bureau of Labor Statistics (BLS). In both cases, the unemployment rate is calculated by taking the projected number people in the civilian labor force and subtracting the projected number of employed individuals, then dividing that number by the projected number of people in the civilian labor force.


While the Federal Reserve posts a set of projections for the unemployment rate, their projections include a variable range. In making their projections, the Federal Reserve uses a range for estimating unemployment across two data sets: the total (Range) and the total minus outliers (Central Tendency). For the Harvard Business Review article, we averaged the 2014 estimate to 6.3%. The Fed Range is 6.2% to 6.7% and the Central Tendency range is 6.3% to 6.6%. For 2016, we averaged the unemployment rate to 5.5%. The Federal Reserve Range was 5.0% to 6.0% and the Central Tendency range was 5.3% to 5.8%.


For purposes of this forecast, we took a business approach to the assumptions within our model. Rather than using purely economic assumptions and leveraging economists, we employed business analysts to help develop an alternative model to the traditional economic rationale. Our projections based on these assumptions suggest numbers toward the low end or below the ranges set by the Federal Reserve.


The following describes the methodology that we used to leverage our proprietary data in forecasting the unemployment rate:


  1. To project the civilian labor force, we use the US Census estimate1 as a proxy for the civilian labor force.  We make no attempt to project and adjust for changes in the labor force participation rate, nor changes to the size of the labor force, other than the natural population growth embedded in the Census estimate.
  2. We use the following information to estimate the change in the number of adults employed: 
    First, we use our survey data2 to forecast the number of new jobs to be added by existing businesses. We obtain this data by directly surveying approximately 3,000 businesses across the United States and asking how many employees they plan to hire in the next 6 months. We leverage that raw data and adjust it to account for business optimism3, 4.
    By calculating historical differences between expectations and actions, tying that to correlating data for optimism and access to capital, we are able to use the survey data to estimate actual hiring based on business owner expectations.
    We use a seasonal adjustment factor to adjust the number of new jobs to be added. The seasonal adjustment factor takes into account normal monthly fluctuations that are repeated annually due to weather and hiring and firing patterns that accompany seasonality and holidays. These seasonal adjustment factors are standard practice, often used by the BLS and other economic institutes when making monthly predictions.
    Then, we employ a forecast for initial jobless claims to estimate the number of job losses in the economy, and also apply a seasonal adjustment factor to that data.5
    Finally, we use data from our proprietary database of businesses across the United States to forecast the addition of jobs that result from new businesses based on past trends.


All these data elements are included in our forecast of net new employees each month. This number is added to the total number of employed persons from the previous month to arrive at a total number of employed persons. This number is then subtracted from a proxy for the civilian labor force, which results in the number of unemployed adults, and then divided by the civilian labor force to obtain the unemployment rate.


See data below:


Unemployment Number Details



  1. US Census Estimate for Adults 18-64.
  2. Survey data is collected in conjunction with Pepperdine University. Businesses are randomly selected and approximately 3,000 businesses, generally distributed geographically and categorically. throughout the US, responded to the survey. The survey results are then extrapolated over the entire universe of businesses in the US. This sample size, while large for a business survey, cannot be held to significance testing across the estimated 25M businesses that currently exist nationally. Nevertheless, we believe there to be strong probabilistic validity and that the data is directionally correct. Additionally, because this data was ascertained through a survey, as with all surveys, the results may be subject to selection bias.
  3. To adjust the survey data for optimism, we use the Conference Board’s Consumer Sentiment Index Present Situation Index to rationalize the survey results. We compare the Index to the base period and use that relative percentage change to adjust the survey results in order to control for outlying business owner sentiment.
  4. For the most conservative projection shown here, we took the low end of the range of jobs projected by business owners over the following six months and extended it across an eighteen-month period. We straight-lined the number evenly across each of the following eighteen months. This assumption, while clearly not likely as a precise monthly outcome, is believed to be conservative due to the typical phase of hiring growth for small businesses. Our analysts have modeled other more aggressive scenarios, which can be made available by contacting our Data Chief, Wisdom Lu at
  5. Initial Jobless Claims forecast is sourced from Trading Economics.