PEW also has an article on this, and they have displaced population data starting in 1951.
The text that goes with it says:
Nearly 1 in 100 people worldwide are now displaced from their homes, the highest share of the world’s population that has been forcibly displaced since the United Nations High Commissioner for Refugees began collecting data on displaced persons in 1951.
Only for the former category (IDPs) data is available starting just from 1989. The large increase since 2011 is due to Syrians. An there's no huge spike in the above graph in 1989 so the inclusion of IDPs doesn't seem to have created a major distortion as Goldstein claims.
Generally speaking, getting IDP data is quite difficult, even for present day conflicts, with sampling being the usual method, because actual accounting is often unfeasible. Quoting from a UNCHR paper on this matter:
In most IDP contexts, no registration systems or other list of the total IDP population exist to enable
analysis on their situation. Where they do exist, they are often incomplete due to the high cost of
maintaining them up to date, as well as various other reasons that will be discussed later in this article.
These gaps, often corresponding to groups likely to be particularly vulnerable, create biases that
should be of concern to the humanitarian community. In the absence of such lists in most countries,
data on the total IDP population is often not available at all, or consist of a patchwork of estimates from
various sources collected for different reasons (most commonly a patchwork of uncoordinated
programmatic assessments implemented by different actors, in different but overlapping places and
time periods). Some examples of including displacement analysis in the national census or large scale
demographic surveys to obtain an overview of the displacement situation exist from some countries,
but in general the opportunity for collecting or analyzing information on internal displacement in these
processes is rarely tapped into.
Hence, most of the time data with complete coverage of the total IDP population is not available.
Instead, a sample-based approach is often used to produce data with the objective of generalizing the
findings to the total IDP population. Choosing a sample should allow for each individual in the IDP
population, be it households or individuals, an equal opportunity to be selected for study. Sample-based
methods are often the most cost-effective and feasible for obtaining needed data on IDPs, and
– done well – they can produce equally useful information for operational decision-making as data
collection with full coverage; a point that is far too often under-valued. Choosing a sample that can
produce representative data on the IDP population requires careful planning based on information, at
a minimum, on the locations and estimates of the size of the overall IDP population, which can be used
as a sampling frame. However, various issues complicate not only the establishment of a
comprehensive sampling frame on IDPs, but also collecting data on the situations of these populations
in general thus further complicating the ability to produce representative data.
There have been some attempts to estimate the number of IDPs for [some] past conflicts published in the World Refugee Survey of the U.S. Committee for Refugees and Immigrants, but these numbers have been called into question by other scholars, e.g. quoting an article published on the World Peace Foundation's website:
Even when relying on a single, supposedly consolidated, source, IDP numbers are likely historically under-reported. Looking at WRS data on Lebanon for 1982-83 i.e. in the middle of the country’s 15-year long civil war, the number of IDPs is placed at 60,000. This figure is hardly believable, given the extensive scale of the fighting since 1975 and that Israel had just invaded the south of the country in mid-1982. In fact, the very next year, WRS’ IDP estimate for the country jumps to 600,000. Similarly, for both Iran and Iraq, WRS estimates of IDPs during the initial years of their bloody 1980-88 war are conspicuously patchy or missing. A frequently cited estimate of the total number of casualties—including those killed and injured—stands at 1 million, for the entire duration of the conflict. Give this figure, one would also expect significant levels of internal displacement to have occurred from the onset along both countries’ border areas. Yet no figures for this are available.
Likewise the same source is skeptical of UHCHR data on IDPs:
Discrepancies in the data become more anomalous when looking at global IDP estimates. A 1991 report commissioned by the UN’s Economic and Social Council, and referenced in later UN documents, estimated the number of IDPs at the time to be 24 million. According to the UNHCR, in 1993–the first year for which the agency has separate IDP figures–the number of IDPs worldwide was just 4.19 million. Let us now assume that both these figures are good approximations. Even if we consider the fact that the UNHCR does not factor in natural disaster-caused displacement, how likely is it that within a span of 2-3 years the number of IDPs actually fell by nearly 20 million? Not very, especially considering the political and often violent turmoil that marked the early 1990s, from Somalia to Afghanistan to the Balkans.
So you can either buy or not Goldstein's argument that the number of displaced would have been higher if those uncounted IDPs from pre-1989 conflicts (partition of India etc.) were properly accounted for. As you can see from papers discussing such issues, it's difficult to get trustworthy estimates for IDPs, and even more so for past conflicts.
On the other hand, the number of refugees (no IDPs included) for 2016 is only about a 1/3 of that total number (22.5 million out of 65.5 million), so only 0.27% or so, comparable in the graph above with the mid-1980s, before IDPs started to be accounted for. So Goldstein is roughly correct on this issue.
Do keep in mind though that stopping would-be refugees from crossing borders in the first place is a growing trend, so the number of people with actual refugee status could be equally a iffy comparison across time if the goal is to use this number estimate something else (like the magnitude of world conflicts or stability of world order, etc.) So Goldstein may not be entirely correct that world order is just as it was decades ago because the number of refugees is just as it was before it was harder to become a refugee. (You haven't quoted this part of his argument, but this why he is ultimately upset by the UN claim, see the finlay "Why it matters" section in his write-up). Quoting from a paper on the spread of recent border restrictions:
While the 1960s and 1970s saw a progressive
expansion of the international refugee protection
regime, from the 1980s onwards refugee policies have
become increasingly restrictive, with industrialised
countries in particular violating international norms
both in the letter and the spirit. Manifestations of
this trend include interdiction, interception, offshore
detention and restrictions to family reunification
rights. These negative attitudes are increasingly
being replicated in lower-income countries that have
hosted large numbers of refugees, often for many
years, and are today home to 85% of the world’s
refugee population. While domestic factors are
clearly at play, it is possible to trace what we term a
‘ripple effect’, with developed countries influencing
each other’s policies and consciously cultivating or
indirectly fostering negative developments in lower income
countries. [...]
Australian policies have undermined refugee protection
in Indonesia in several ways. First and foremost,
refugees’ prospects for resettlement have diminished
as Australia has cut resettlement quotas, in a context
where Indonesia continues to reject local integration.
Australian policy has led to greater restrictions on
the part of Indonesia, including the increased use of
immigration detention, the growing criminalisation of
refugees and boat pushbacks. All of these measures
are in clear contradiction of the spirit of the Refugee
Convention. [...]
Jordan’s sealed borders effectively leave refugees
fleeing conflict in Syria with three main options:
remaining in Syria, moving into Jordan irregularly
or seeking refuge elsewhere, with Europe one major
option. To the extent that Jordan’s policies may
prompt the first two of these options, there are clear
concerns in terms of refugee protection and rights
under the 1951 Refugee Convention. However, they
could also prompt increased flows to Europe. In
the latter case, there is a sense in which European
restrictions aimed at reducing flows to Europe have
in fact encouraged a Jordanian approach that may in
the long term increase the number of Syrians arriving
at Europe’s borders.
