In this blog post, we’ll explore the significance of forecasting future urban populations—a critical aspect of urban planning—as well as the characteristics and applications of key population forecasting models, such as the trend extrapolation method, the comparative inference method, and the generation method.
If a piece of clothing is too small, it will be tight and uncomfortable; if it is too large, there will be excess fabric relative to the body, reducing efficiency. The same applies to cities: if a city is too small relative to its population, it becomes inconvenient to use; if it is too large, the city’s efficiency declines. Urban designers plan cities and determine the scale of various infrastructure facilities by taking future urban populations into account. The ultimate goal of urban planning is to expand infrastructure to prepare for future urban growth and to manage that growth efficiently. For this reason, population forecasting is extremely important in the field of urban planning. Since cities come in various forms—ranging from existing cities to new towns—it is impossible to use a single method for population forecasting. Therefore, a variety of methods are needed to complement one another. Generally, the estimated population 20 to 30 years in the future is referred to as the planned population. The primary methods for forecasting the planned population include the trend extrapolation method, the comparative inference method, and the development method.
The trend extrapolation method forecasts the planned population by extending past population trends into the future, based on the assumption that future population trends will resemble those of the past. This method is useful when no significant population inflow is expected during the projection period and the city is anticipated to grow at a rate similar to past trends. The most basic form of the trend extension method is the linear model. The linear model is suitable when the population tends to increase or decrease by the same absolute amount each year; it predicts the projected population using the average annual rate of population growth or decline. For example, if the current population of Osan City is 200,000 and an analysis of past population statistics shows that it has increased by 10,000 people annually, the population of Osan City 20 years from now can be predicted using the linear model as follows: 200,000 + 10,000/year × 20 years = 400,000.
While the linear model deals with “numbers,” the exponential model deals with “rates.” The exponential model is used when a city’s population tends to increase or decrease at a constant rate each year. Like the linear model, the exponential model has the advantage of relatively simple calculations; however, it has the limitation that assuming the same growth or decline rate will persist over a long period can result in significant discrepancies from reality. For example, if an analysis of past population statistics shows that Gimhae City’s current population is 500,000 and the city’s population has increased by 1% annually, the population of Gimhae City 20 years from now can be predicted as 500,000 × (1 + 0.01)²⁰ ≈ 610,000.
To compensate for the limitations of the simple exponential model, a logistic model can be used that accounts for the long-term slowdown in population growth. In urbanizing cities, the planned population may increase at a relatively rapid rate initially, but as time passes, the growth rate gradually slows down, ultimately converging toward a steady state—a pattern known as an S-shaped population growth curve. The logistic model accounts for this S-shaped population growth curve and, unlike the two previous models, is well-suited to reflecting the long-term deceleration of population growth. However, as noted earlier, simply applying different growth rates—such as an average annual rate of 5% initially and 1% thereafter—is closer to a stepwise application of exponential growth rates than a strictly defined logistic model. Therefore, when actually applying the logistic model, the S-shaped growth curve must be calculated by considering factors such as the upper limit to which the population is expected to converge in the long term and the growth rate.
The trend-extension method has the advantage of allowing urban population projections through simple calculations. However, there are cases where it is difficult to predict future population based solely on past population trends—either because sufficient data on population changes is unavailable or because social factors affecting population are expected to change rapidly. To overcome these limitations, the comparative inference method is used; this method involves inferring population changes in the target city based on the population growth patterns of other cities. Representative comparative inference methods include the ratio method and the comparative method.
The ratio method involves predicting population changes in the target city based on population changes in a broader region that includes the city. For example, when attempting to forecast the population of Gokseong but insufficient data on Gokseong’s population changes is available, this method uses population change data from South Jeolla Province—the region to which Gokseong belongs—to predict Gokseong’s population changes. If Gokseong currently accounts for 5% of the total population of Jeollanam-do, and the population of Jeollanam-do is projected to reach 2 million in 20 years, then using the ratio method, we can predict that Gokseong’s population will be 5% of 2 million, or 100,000. Thus, to use the ratio method, it is necessary to assume a “constant share”—that is, that Gokseong’s population ratio to that of South Jeolla Province (the higher-level region) will remain constant in the future.
Another method of comparative inference is the comparative method, which involves selecting a specific region deemed most similar to the city being projected in terms of physical and socioeconomic characteristics, and then inferring the population change of the target city based on the population changes in that region. This method is primarily used for cities where rapid social demographic shifts are anticipated. Since this method relies on the city most similar in nature for population forecasting, identifying such a city is the most critical factor in determining the accuracy of planned population forecasts.
Both the past trend extrapolation method and the comparative analogy method examined earlier predict only changes in total population size; therefore, they do not provide a sufficient answer to the question of how the projected population is composed. Consequently, the composition method is used to supplement these approaches. The composition method is a technique for measuring a city’s population change based on its constituent elements. By analyzing each component of the population—births, deaths, and population inflows and outflows—the composition method allows us to determine population changes by age group. In the component method, women of childbearing age are typically defined as those aged 15–44; the number of births can be estimated by considering the size of the female population in this age group, along with the fertility rate and sex ratio. Additionally, the method takes into account both the process by which the existing population moves into the next age group as it ages and the inflow and outflow of the population. In this way, the component method allows us to determine not only the total size of the urban population but also its composition.
Various methods are used to accurately forecast projected population figures. However, since all of the above methods are based on demographic data, highly accurate statistical data is a prerequisite. Based on the size and composition of the projected population derived from accurate statistical data, urban planners will be able to design a city that fits people like a glove—that is, a city that everyone can use comfortably.
