Plans Under Pressure: Expectations, Uncertainty, and Hiring Reality

A potential employer can have the intent, capital, and customers associated with employee number one and still miss the transition because the plan was built on uncertain information. This chapter uses application-stage hiring forecasts to examine that problem: what owners expected, what actually happened, and why the gap appears in both entry and employment scale. The evidence supports a careful diagnosis—expectations contain useful information, but they are noisy, selected, and vulnerable to local and macroeconomic shocks.

A hiring forecast is evidence, not a promise

The business-formation study follows about 17 million applications filed between April 2017 and December 2021. These are applications for Employer Identification Numbers that state an intention to start a new business. The application asks for the highest number of employees the applicant expects during the next twelve months. A positive number is treated as expected employer entry; zero is treated as no expected employer employment.

The study then links applications to later employment records. Realized entry means that the linked business records positive employment in the four quarters after the application, with a timing adjustment for the week in which the application was filed. Maximum realized employment is the highest quarterly employment observed during that window. The comparison is therefore between a plan made near formation and an outcome measured over a defined twelve-month horizon.

That distinction is important. An expectation is not a hiring contract. It does not require the applicant to hire, and it can be revised as demand, financing, regulations, or personal circumstances change. The form is also primarily an administrative document rather than a high-stakes investor forecast. That makes the response a useful window into what applicants believed at the outset, while leaving open how carefully every number was calculated.

The data contain a particularly informative contrast. Applications with a planned first wage payment have a 27% employer-transition rate within four quarters, compared with 7% for all applications in the sample. Planned-wage applications almost always report positive expected employment. A planned wage date therefore contains information about concrete hiring intent, but it also selects a more committed group. It would be wrong to apply the 27% rate to every person who files an application, just as it would be wrong to treat every zero-expectation application as a failed employer.

Many zero-expectation applications may represent intended nonemployer businesses or delayed hiring. For the study’s employment-generation outcome, those businesses are recorded as producing zero employment, even if they later provide income to their owners. This is a sensible definition for estimating employer entry. It is not a measure of whether the underlying business formed, earned money, or improved the owner’s alternatives.

The forecast also measures two different margins. The extensive margin is whether the business becomes an employer at all. The intensive margin is how many employees it has conditional on becoming an employer. A forecast can fail because entry never occurs, or because entry occurs but the business hires fewer or more workers than expected. Treating both events as one “bad forecast” hides the mechanics of the first employee.

Why the average is wrong in two directions

The results show aggregate overestimation, but the average conceals a sharp split between people who do not enter and people who enter at a different scale. In the full application sample, 84.1% neither expected nor achieved employer entry. Another 8.9% expected entry but did not realize it. A further 3.6% entered despite not expecting to, and 3.4% both expected and achieved entry.

Across all applications, expected employment averaged 0.530 employees, while realized employment averaged 0.388. The average expectation error—expected maximum employment minus realized maximum employment—was therefore positive 0.142 employees. On its face, applicants overpredicted. But that result is driven substantially by people who forecast employer entry and then did not enter.